EI2024 Paper-Abstract Searchable

Paper
TITLE First Author Conference

Number
JIST-first: Camera Motion Estimation

3DIA-100 Method Using Depth-Normalized Seok Lee 3DIA
Criterion
A VGA Light-Tracing Time-of-Flight CMOS

3DIA-101 Image Sensor with Multi-Tap Pixel for Kamel Mars 3DIA
Long-Range Outdoor Operations
Extending Lidar Depth Range Using

3DIA-102 Filip Taneski 3DIA
Intensity Data
A Deep Learning Based Light Field Image

Soheib
3DIA-103 Compression as Pseudo Video Sequences 3DIA
Takhtardeshir
with Additional in-loop Filtering
Data Augmentation Based on Depth Hamed Razavi

3DIA-104 3DIA
Information for Neural Radiance Fields Khosroshahi
N-DEPTH: Neural Depth Encoding for Stephen

3DIA-105 3DIA
Compression-Resilient 3D Transmission Siemonsma
Evaluation of Signal and Noise Metrics of

AVM-107 High Dynamic Range Image Sensors by Orit Skorka AVM
IEEE P2020 Methodology
Measurement of Noise Equivalent

AVM-108 Quanta (NEQ) Using the Deadleaves Robin Jenkin AVM
Technique
Measuring Natural Scenes SFR of
AVM-109 Daniel Jakab AVM
Automotive Fisheye Cameras
An Integrated and Fully Automated Test

AVM-110 Solution for Automotive Cameras of Xingbo Wang AVM
Ultrawide Angle of View
Multi-Modal Pedestrian Detection via

Dual-Regressor and Object-Based Napat
AVM-111 AVM
Training for One-Stage Object Detection Wanchaitanawong
Network
Exploring Object Detection and Image

AVM-112 Classification Tasks for Niche Use Case in Ryan Peruski AVM
Naturalistic Driving Studies
Efficient Fault Tolerant Architecture for Shyam

AVM-113 AVM
Neural Network Compute Jagannathan
Fusing Segmentation and Domain-

knowledge Model to Extract Intersection
AVM-114 Julien Vijverberg AVM
Topologies from Aerial/terrestrial
Orthographic Images
Optimizing Ego Vehicle Trajectory

AVM-115 Prediction: The Graph Enhancement Sushil Sharma AVM
Approach
An Approach to Recreate Large Virtual

AVM-116 Environments for Use in Road Traffic Jack Miller AVM
Research
Urban Scene Reconstruction based on

AVM-365 Shen Shihao AVM
Neural Rendering
KEYNOTE: Image Restoration through
COIMG-117 Mauricio Delbracio COIMG
Inversion by Direct Iteration (InDI)
Provable Probabilistic Imaging Using

COIMG-118 Yu Sun COIMG
Score-Based Generative Priors
Transformers for Microscopy Slide Image

COIMG-119 Franklin Wang COIMG
Segmentation of Invasive Melanoma
Generative Plug-and-Play: Posterior

COIMG-120 Gregery T. Buzzard COIMG
Sampling for Inverse Problems
Keynote: Efficient Neural Scene

COIMG-121 Representation, Rendering, and Gordon Wetzstein COIMG
Generation
COIMG-122 Dark Secrets of Computational Imaging Laura Waller COIMG
Super Resolution in Heterogeneous Array

Cameras under Coherent and Partially Kevin Chew
COIMG-123 COIMG
Coherent Illumination: A Generative AI Figueroa
Approach
Diffractive Processing of Visual

COIMG-124 Aydogan Ozcan COIMG
Information
Physics-informed Generative Models for Muhammad

COIMG-125 COIMG
Imaging and Photonic Design Tayyab
Considerations for Materials Design from

COIMG-126 Jeff Simmons COIMG
the Latent Space
Methodology for Estimation of Intrinsic

COIMG-127 Dimensions and State Variables of Megna Shah COIMG
Microstructures
Persistent Homology for Microstructure

COIMG-128 Simon A. Mason COIMG
Manifold Construction
Dissipative Lagrangian Neural Networks Veera
COIMG-129 COIMG
for Materials Imaging and Simulation Sundararaghavan
Mathematical Nuances of Gaussian-

Marcus Michael
COIMG-130 Process-Driven Autonomous COIMG
Noack
Experimentation
Reinforcement Learning for Metals

COIMG-131 Ojash Neopane COIMG
Powder Bed Fusion
Evaluation of Information Content in 2D

COIMG-132 and 3D Microstructural Characterization Ayesha Abdullah COIMG
of Brush Particle-Based Hybrid Materials
Semantic Segmentation and Anomaly

COIMG-133 Identification of Binder Jet Powder Bed Alexander Gourley COIMG
Images
COIMG Keynote: Quantitative Secondary

COIMG-134 Electron Yield Mapping in Ion-beam Akshay Agarwal COIMG
Microscopy
Biomedical Ultrasound Computed

COIMG-135 Tomography Using Neural Full Waveform Zhijun Zen COIMG
Inversion
Automated, High-Throughput X-ray CT

Characterization of Additively Amir Koushyar
COIMG-136 COIMG
Manufactured Parts Using Deep Learning Ziabari
Based Algorithms
COIMG KEYNOTE: Understanding

Microstructural Evolution in Additively
COIMG-137 Daniel Sinclair COIMG
Manufactured Materials by 3D X-ray
Imaging
Multimodal Registration and Fusion of In

COIMG-138 Situ and Ex Situ Process Monitoring Data: Sean Donegan COIMG
A Case Study in Laser Powder Bed Fusion
Engineering Temporally Integrated
Images for In Situ Monitoring and
COIMG-139 Zackary Snow COIMG
Analysis of Laser-Based Powder Bed
Fusion Additive Manufacturing
Benefit of X-ray Imaging to Improve Anne-Françoise

COIMG-140 COIMG
Additive Manufacturing Obaton
Model-based Deep Learning with a CNN

Prior for Sparse-view Cone-beam CT
COIMG-141 Aniket Pramanik COIMG
Reconstruction of Additively
Manufactured Objects
Leveraging Pixel Value Certainty in Pixel-

COIMG-142 Shift and Other Multi-Shot Super- Henry G Dietz COIMG
Resolution Processing
Multimodal Deep Learning Approach for

Dynamic Sampling with Automatic
COIMG-143 Feature Selection in Matrix-Assisted Laser David Helminiak COIMG
Desorption/Ionization Mass
Spectrometry Imaging
Gradient-Based Image Visualization for

COIMG-144 Taichi Nakano COIMG
Noisy LWIR Image
COIMG KEYNOTE: Breaking Boundaries in

AM: The Dynamic Duo of Process
COIMG-145 Ankit Saharan COIMG
Development and Closed Loop Control in
Laser Powder Bed Fusion
Generative Denoising Models for Patxi Fernandez-

COIMG-147 COIMG
Material Science Applications Zelaia
On the Development of Acoustic Melt
COIMG-148 Pool Monitoring Techniques Assisted by David Stobbe COIMG
High-speed Synchrotron X-rays
Classification of Melt Pool Boundaries

and Defects through Dual-energy X-ray
COIMG-149 Obaidullah Rahman COIMG
CT of Crept Additively Manufactured
Parts
JIST-first : Computational Image

COIMG-150 Formation: Simulators in the Deep Stanley Chan COIMG
Learning Era
Beyond RISC - Next-generation Compute

COIMG-151 Sergio Goma COIMG
Architecture for Computational Imaging
COIMG-152 PSF Engineering for Edge Sensing Nicholas Antipa COIMG
Reconstruction of Phase and

COIMG-153 Temperature Maps from Energy-resolved Anton S. Tremsin COIMG
Neutron Imaging
Bio-Inspired Methods for Efficient Visual
COIMG-154 Emma Alexander COIMG
Sensing
COIMG-155 Mobile Aware Computational Imaging Jing Li COIMG
Accurate Efficient Multispectral

COIMG-156 Spatiotemporal Image Reconstruction Luke Lozenski COIMG
Using Neural Field Representations
Neural Deconvolution and Rendering for

COIMG-157 Suren Jayasuriya COIMG
Synthetic Aperture Sonar
Implicit Neural Representation for

COIMG-158 Yu Sun COIMG
Tomographic Imaging
Implicit Representation for Motion-
COIMG-159 Kunal Gupta COIMG
Resolved Computed Tomography
Maximum Likelihood-based Phase-

Retrieval Using Fresnel Propagation
COIMG-160 K. Aditya Mohan COIMG
Forward Models with Optional
Constraints
Detection of Multiple Fluorescent Dyes

COIMG-161 Using Liquid-crystal Variable Retarder Kazuma Fujiwara COIMG
and Sparse Modeling
Open Set Domain Adaptation for Image

Classification with Multiple Unknown
COIMG-162 Daichi Nishihara COIMG
Labels Using Unsupervised Clustering in a
Target Domain
3D Printing Images Defect Detection and

COIMG-368 Classification Supported by Generative AI Amra Peles COIMG
Models
3D Printing Images Defect Detection and

COIMG-368 Classification Supported by Generative AI Amra Peles COIMG
Models
Experimental Study on Reducing

Wavelength Dependency of Spatial
COLOR-163 Tsubasa Ando COLOR
Resolution Characteristics of a Digital
Camera by MTF Correction
Double Binnable LMS and RGBW Color

COLOR-164 Mritunjay Singh COLOR
Filter Arrays
Joni Mikael
COLOR-165 Camera Color Correction Using Splines COLOR
Suominen
Color Calibration Methods for OLED

COLOR-166 Maliha Ashraf COLOR
Displays
Characterisation of Color Enhancing

COLOR-167 Screen Printing Technique with RGBW Alina Pranovich COLOR
Interference Pigments
Equalization of Appearance Using

COLOR-168 Che Shen COLOR
Individualized Unique Hues
A Method for Estimating Spectral

Reflectance from Camera Responses
COLOR-169 Shoji Tominaga COLOR
Using Local Optimal Dataset and Neural
Network
Investigating Consistency of Color

COLOR-170 Appearance Using Different Reproduction Apurva Zunjarrao COLOR
Approaches
Visual Assimilation Illusions: Perceptions

COLOR-171 John J. McCann COLOR
or Optics?
JIST-first: Spatio-Temporal Retinex-

COLOR-172 Inspired Envelopes with Anisotropic Peter Sagvold COLOR
Diffusion
JIST-first: Spectral Estimation: Its

COLOR-173 Behaviour as a SAT and Implementation Tanzima Habib COLOR
in Colour Management
Computational Estimation of Head Pose

CVAA-174 and Gender Through Large Corpora of Ali Unlu CVAA
Fine-art Portraits
TextAIles: Color Manipulations in the

CVAA-175 Ludovica Schaerf CVAA
Latent Space of StyleGAN3
Distant Viewing Light Analysis of Pepe Ballesteros
CVAA-176 CVAA
Figurative Paintings Using Spherical Zapata
Towards Artist Recognition Based on

CVAA-177 Material Rendering. A Case Study for Jing Huang CVAA
Recognition of Rembrandt and Van Dyck
The Use of Transfer Learning for the

CVAA-178 Hassan Ugail CVAA
Analysis of Paintings by Raphael
Vision Transformers and Synthetic Image

CVAA-179 Ludovica Schaerf CVAA
Augmentation in Art
JIST-first: Spatial Analysis and Visual

ERVR-180 Communication of Emergency Sharad Sharma ERVR
Information through Augmented Reality
Real-time Stereoscopic Image-parallel

ERVR-181 Erwan Leria ERVR
Path Tracing
Interactive Aviation Maintenance LaTasha Taylor

ERVR-182 ERVR
Classroom Starr
HPCI KEYNOTE: High Performance

HPCI-183 Imaging Applications: At the Intersection Mohamed Wahib HPCI
of HPC and AI
High-performance Data Management for

HPCI-184 Haoju Leng HPCI
Computational Pathology
Alleviating Tiling Effect by Random Walk

HPCI-185 Sliding Window in High-resolution Shunxing Bao HPCI
Histological Whole Slide Image Synthesis
Scalable Analysis of Giga-pixel WSI
HPCI-186 Sajal Dash HPCI
Images
Back-Projection Optimization: The Key to

HPCI-187 Peng Chen HPCI
Tomographic Imaging
Accelerated Laminographic Image

HPCI-188 Bin Ma HPCI
Reconstruction Using GPUs
Focal Talk: A Visual Cortex for Array

HPCI-189 David Brady HPCI
Cameras
Focal Talk: The Image Systems

HPCI-190 Brian Wandell HPCI
Engineering Toolbox
Taking Pictures of Planetary Interiors

HPCI-191 Tolulope Olugboji HPCI
using Powerful Computers
HPCI KEYNOTE: Scientific Machine

Learning for Computational Wave
HPCI-192 Youzuo Lin HPCI
Imaging Problems: from Carbon Zero
Emissions to Breast Cancer Detection
Automatic Optimization for Fast

HPCI-193 JiangChen HPCI
Convolution for SIMD Architecture
JIST-first: Digital Modeling on Large

HPCI-194 Quan Liu HPCI
Kernel Metamaterial Neural Network
Facilitating Efficient Scientific Data
HPCI-195 Streams through High-End Data Transfer Tekin Bicer HPCI
Nodes
High-performance Data Loader for Large- Edgar Josafat

HPCI-196 HPCI
scale Data Processing Martinez-Noriega
Asynchronous Tensor Decomposition

HPCI-197 Applied in High Performance Image Zhen Zhou HPCI
Processing
Rapid and Scalable Training for Ultra-

HPCI-198 Long Sequence Vision Transformer for Xiao Wang HPCI
Imaging Applications
Efficient Distributed Sequence Parallelism

HPCI-199 for Transformer-based Image Isaac Lyngaas HPCI
Segmentation
KEYNOTE: S(t)imulating Art and Science:

HVEI-200 Marton Orosz HVEI
György Kepes and the concept of
Beyond Space and Time: Analysis of the

Arcane Perceptual & Cognitive Effects in
HVEI-201 Lora Likova HVEI
Ivan Vukadinov’s Art at the Vatican
Museums
Michelangelo's Cryptic Self-Portraiture in
HVEI-202 Christopher Tyler HVEI
the Sistine Chapel
Categorizing Art Viewing Behavior Using

HVEI-203 Elena Fedorovskaya HVEI
Fuzzy Logic
You're in the Picture: Revisiting the

HVEI-204 Michael Kubovy HVEI
Robustness of Perspective
Surreal Artist as Visual Neuroscientist:

HVEI-205 Analysis of Selected Works by Rene Russell D. Hamer HVEI
Magritte
Entropy Ordered Shapes as Bivariate

HVEI-206 Nicolas S. Holliman HVEI
Glyphs
Perception of Tactile Attributes of

HVEI-208 Alireza Rabbanifar HVEI
Colored Texture Images
Spatiotemporal Contrast Sensitivity

HVEI-209 Functions: Predictions for the Critical Ali Bozorgian HVEI
Flicker Frequency
HVEI KEYNOTE: Design and Reality of the

HVEI-210 Robin Atkins HVEI
HDR Imaging Ecosystem
JPI-first: The Impact of Adaptation Time
HVEI-211 in High Dynamic Range Luminance Jake Zuena HVEI
Transitions
Modeling Lightness and Color Perception

HVEI-212 Michael Rudd HVEI
of High Dynamic Range Stimuli
Perception in Mixed Reality

HVEI-213 Bernice Rogowitz HVEI
Environments
Egocentric Boundaries on Distinguishing

HVEI-214 Colliding and Non-Colliding Pedestrians Alex D. Hwang HVEI
while Walking in a Virtual Environment
Digital Rear-view Mirrors with

HVEI-215 Augmented Reality in Comparison with Miao Zhang HVEI
Traditional Rear-view Mirrors
Study on the Relationship Between

HVEI-216 Quality and Acceptability Annoyance Pierre Lebreton HVEI
(AccAnn) of UGC Videos
Comparison of Subjective Methodologies

for Local Perception of Distortion in
HVEI-217 Andréas Pastor HVEI
Videos and Impact on Objective Metrics
Resolving Power
Advancing Video-Wise Just Noticeable

Difference (VW-JND): Innovative
Strategies for Dataset Collection,
HVEI-218 Jingwen Zhu HVEI
Procedure Optimization, and
Comparative Analysis of Video Quality
Metrics
Metric Thresholds for Acceptance and
HVEI-219 Annoyance of User Generated Video Ali Ak HVEI
Content
HVEI KEYNOTE: Quality Assessment in

HVEI-220 Context: The QoE Way Towards Patrick Le Callet HVEI
Sustainable Media Consumption
Color Perception-Driven Energy

HVEI-221 Optimization for Virtual Reality: Displays Yuhao Zhu HVEI
and Image Encoding
Study on Effect of Display Layout for Web

HVEI-222 Conferencing Services on Subjective Kimiko Kawashima HVEI
Evaluation Stability
Background- and Ambient-aware Image

HVEI-223 Visibility Enhancement for Transparent Seungchul Ryu HVEI
Displays
Adapting Pretrained Networks for Image

HVEI-224 Quality Assessment on High Dynamic Andrei Chubarau HVEI
Range Displays
Image Quality Assessment across Viewing

HVEI-225 Distances: A Comparison Study of CSF- Dounia Hammou HVEI
based and Rescaling-based Metrics
HVEI KEYNOTE: From Neurons to Pixels,

HVEI-226 Anjul Patney HVEI
to Neurons Again
Focal Talk: Comprehensive Insights into

HVEI-227 Psychological Image Quality: Addressing Mikolaj Leszczuk HVEI
Ecological Validity and Emotional Content
HVEI-228 One Number to Describe Them All Lucjan Janowski HVEI
Aesthetics and Entropy. IV. Image

HVEI-229 M.R. V. Sahyun HVEI
Composition and Content
Analysis of User Preferences for Film
HVEI-230 Grain Noise in Images and Video Xinyi Guo HVEI
Sequences
JPI-first: Mid-level Characteristics of

HVEI-231 Drawings Made by Observers with Benjamin Balas HVEI
Aphantasia
Towards Enhancing Ecological Validity in

HVEI-232 Sruti Subramania HVEI
QoE User
How AI Image Synthesis Changes the Role

HVEI-233 Scott Daly HVEI
of Vision Science in Imaging Applications
Perceptual Experiment for Searching the

HVEI-234 Spatial Frequency Threshold Depending Dong-Yeol Yeom HVEI
on Pixel Design
Effect of Brightness and Size on Display

HVEI-235 Flicker Perception: Comparison with Hyosun Kim HVEI
Flicker Indices
Heaven on Earth Map: “As Above, So

HVEI-236 Bradly Couch HVEI
Below.”
User Performance and Quality of

HVEI-237 Experience for a Remote-controlled Lab- Kawthar El Ouardi HVEI
based Moving Platform
IMAGE-238 Adapt to Distill or Distill to Adapt Georgi Thomas IMAGE

Deep Learning-Based Defect Detection
IMAGE-239 and Segmentation for Wood Plank Mazhar Mohsin IMAGE
Surfaces with Real-Time Tracking
Motion-based Domain Randomization for

IMAGE-240 Chih-Hsing Ho IMAGE
Detecting Honeybees Inside a Hive
Self-Attention Enhanced Recognition: A

Unified Model for Handwriting and
IMAGE-241 Gaurav Patel IMAGE
Scene-Text Recognition with Improved
Inference
Generalizing Handwriting and Scene-Text

IMAGE-242 Taewook Kim IMAGE
Detection in Images
Neuro-Symbolic Ensembles for Assistance

IMAGE-243 Bob Price IMAGE
at the Edge
Vempuluru
X-ray Image Denoising for Baggage
IPAS-244 Venkata Poorna IPAS
Screening Using Learning Based Methods
Sekhar
Image Restoration via Collaborative

IPAS-245 Wenzhu Xing IPAS
Filtering and Deep Learning
Towards Realistic Landmark-Guided Fatemeh Ghorbani

IPAS-246 IPAS
Facial Video Inpainting Based on GANs Lohesara
Examining the Effects of Compression on

IPAS-248 Deep Learning Remote Benjamin Sporrer IPAS
Photoplethysmography
Coverage Optimization for Training Data

IPAS-249 Raegeun Par IPAS
enhancement of Ill-performing Classes
IPAS-250 Lensless Phase Microscopy Imaging Igor Shevkunov IPAS
Spectral Rendering: From Input to

IPAS-251 Ayoung Kim IPAS
Rendering Process
Automatic Reading Order Sequencing: A

Md Manzoor
IPAS-252 Novel Reading Order Generator for Text- IPAS
Murshed
Based Documents
Data and Label Graph Fusion for Semi-

IPAS-253 supervised Learning: Application to A. Baradaaji IPAS
Image Categorization
Novel Image Super-Resolution and

IPAS-254 Denoising Using Implicit Tensor-Product János Horváth IPAS
B-Spline
Scalable and Efficient Orchestration of

IPAS-255 Machine Learning Workloads on DSPs Aaron Sequeira IPAS
with Multi-level Memory Architecture
Depth-of-Field Guided Rendering for

IPAS-367 Kamran Akbar IPAS
Light-Field Displays
KEYNOTE: Image Information Metrics

from Slanted Edges: a Ttoolkit of Metrics
IQSP-256 Norman Koren IQSP
to Aid Object Recognition, Machine
Vision, and Artificial Intelligence Systems
ISP tuning for Improved Image Quality in

IQSP-257 Diarmaid Geever IQSP
Machine Vision
Resolution and Noise Trade-off Analysis

IQSP-258 Weijian Zhang IQSP
in Single-Photon Pulse LiDAR
Benchmarking Motion Blur of Video-
IQSP-259 Frame-Interpolated Hybrid EVS+CIS Kamal Rana IQSP
against conventional CIS
Evaluating Camera Performance in Face-

IQSP-260 Megan Borek IQSP
Present Scenes with Diverse Skin Tones
From Video Conferences to DSLRs: An In-

IQSP-261 depth Texture Evaluation with Realistic Daniela Carfora IQSP
Mannequins
Focal Talk: Project Paris Insight:

IQSP-262 Hervé Macudzinski IQSP
Methodology
A Novel Deep Learning-Based Anomaly

IQSP-263 Detection Pipeline Optimized for Dark Dayun Lee IQSP
Blemish Pattern of Image Sensor
No-Reference Color Dot Artifact

IQSP-265 Subin Han IQSP
Assessment for Remosaiced Images
Focal Talk: Android Camera Apps:

IQSP-266 Designing in a Sea of Thousands of Clemenz Portmann IQSP
Cameras
Quality of Experience of Virtual Reality-

IQSP-267 Mylene Farias IQSP
Based Communication Applications
Evaluation of Subjective Video Quality of

IQSP-268 Eddie Pei IQSP
Television Displays
Quantified Impact of Key System

IQSP-269 Parameters on the Diffraction Quality in Soyoung Kwon IQSP
Displays
Lens Flare Measurement for RGB Black-
IQSP-270 Laurent Chanas IQSP
box Camera
Recommendations for the Detection and

IQSP-271 Analysis of the ISO 12233:2023 e-SFR Sarah Kerr IQSP
Slanted Star
Advancing Hyperspectral Imaging

Siavash
IQSP-272 Endoscopes (HySE): A Comprehensive IQSP
Mazdeyasna
Performance Evaluation Approach
Luminance and Color Accuracy of

Prosumer and Mobile Phone Cameras for
IQSP-273 Leah Humenuck IQSP
High Dynamic Range Imaging in Cultural
Heritage Documentation
Assessment of Dynamic Range and Noise

IQSP-274 Seungwan Jeon IQSP
in a Natural-Scene Image
Objective Color Characterization of HDR

Videos Captured by Smartphones:
IQSP-275 Anna Tigranyan IQSP
Laboratory Setups and Analysis
Framework
Synthetic Data Generation for AI-based

IRIACV-276 Frederik Seiler IRIACV
Machine Vision Applications
Unsupervised PCB Anomaly

IRIACV-277 Chih-Hui Ho IRIACV
Segmentation with Foundational Models
Real-time Cattle Intake Monitoring Using

IRIACV-278 Prajwal Rao IRIACV
Stereo Vision
IRIACV-279 Mobile 3D Mapping of Erdstall Facilities Raimund Edlinger IRIACV

Comparison of SFR and Edge Artifacts
ISS-280 Between Single Photodiode and Split Xunzhi Li ISS
Photodiode Pixels in HDR Image Sensors
Modeling and Reducing Energy

ISS-281 Consumption of Computational Image Yuhao Zhu ISS
Sensors: A Case Study on Gaze Tracking
Single Slope Single Ramp Column ADC Pedro Nuno T.

ISS-282 ISS
with Digital CDS for CMOS Image Sensors Santos
Intracell Capacitances in Pixel-Level

ISS-283 Valery Axelrad ISS
Hybrid BondImage Sensor Cells
ISS KEYNOTE: Passing the Visual Turing

ISS-284 Grace Kuo ISS
Test with Holographic Displays
Adaptive Bit Depth Control for Neural

ISS-285 Youngil Seo ISS
Network Quantization
Joint Design of Camera Spectral

ISS-286 Sensitivity and Color Correction Matrix Xinyue Yu ISS
with Noise Consideration
Simulation of Indirect Time of Flight Pixel

Bruno Henrique
ISS-287 Using High-frequency Operation Low- ISS
Lopes
voltage Trench Gates
A Dynamic-Range Extension Technique

ISS-288 with Linear-Logarithmic Response for Tomohiro Okuyama ISS
Short-Pulse Time-of-Flight Image Sensors
Joint Parameter Estimation for Event-

ISS-289 Xiaozheng Mou ISS
based Vision Sensor Characterization
Dynamic Low Bit-Depth Quanta Video
ISS-290 Prateek Chennuri ISS
Imaging with Recurrent Restoration
Magic Lantern as a Platform for Digital Paul Selegue

ISS-292 ISS
Photography Research Eberhart
Probabilistic and Deterministic Inversions

MLSI-293 Balu Nadiga MLSI
of Radiographic Data
Neural Network-based Analysis of

MLSI-294 Polychromatic PSFs for Wavelength- Jeongyong Shin MLSI
Specific Extraction
Density Reconstruction from Noisy

MLSI-295 Radiographs Using an Attention-based Daniel Serino MLSI
Transformer Network
Parameter Estimation Using Physics-

MLSI-296 based Statistical Machine Learning for Soumi De MLSI
Radiography
Scatter Removal in Dynamic X-Ray
MLSI-297 Siddhant Gautam MLSI
Tomography Using Robust Features
Supervised Reconstruction for Silhouette

MLSI-298 Evan Bell MLSI
Tomography
Breast Ultrasound Report Generation

MLSI-299 Jaeyoung Huh MLSI
using LangChain
Swap-Net: A Memory-Efficient 2.5D

MLSI-300 Cascade Network for 3D Image Xiaojian Xu MLSI
Reconstruction
Using Neural Networks for Compact

Representation of Features with Singular
MLSI-301 Oleg Korobkin MLSI
Points in Projections of Surfaces on
Radiographic Plane
Local Entropy Key Points for Materials
MLSI-302 Zachary Varley MLSI
Science Image Collation
Combining Multimodal Fatigue Fracture

MLSI-303 Katelyn Jones MLSI
Surface Image for Analysis with a CNN
A Lightweight Transformer for Faster and

MLSI-304 Harry Dong MLSI
Robust EBSD Data Collection
Recovering a Molecule's 3D Dynamics

MLSI-305 from Liquid-phase Electron Microscopy Enze Ye MLSI
Movies
A Fully Differentiable Hydrodynamics

MLSI-306 Framework: Challenging Traditional ML Hyun Lim MLSI
for Optimization in Hydrodynamics
Unpaired Deep Learning for

MLSI-307 Pharmacokinetic Parameter Estimation Gyutaek Oh MLSI
from Dynamic Contrast-Enhanced MRI
Applying Machine Learning to Detect

MLSI-308 Sean Tyler Tronsen MLSI
Anomalies in Additive Manufacturing
What’s Wrong with End-to-end Deep

MLSI-309 Wenjie Zhang MLSI
Learning Methods for Phase Retrieval?
Segmentation of Starch Granules in

MLSI-310 Ye Jin MLSI
Microscopic Images Using a U-Net Model
MOBMU-311 ACES Workflows in Epic Unreal 5 Eberhard Hasche MOBMU
Color Calibration for Multicolored

MOBMU-312 Simon Bugert MOBMU
Barcodes Using Smartphones
Generating Rich Media Posters for

MOBMU-313 Interactive Hiking Trails to Promote Eberhard Hasche MOBMU
Sustainable Tourism
Driver Monitoring System Using Deep Moustafa Elhassan

MOBMU-314 MOBMU
Learning Techniques Ibrahim
Impact of Elon Musk and Bill Gates
MOBMU-315 Tweets on Tesla and Microsoft Stock Maiia Kochneva MOBMU
Prices
Prototypical Implementation of an Susanne

MOBMU-316 MOBMU
Accessible Vocabulary Trainer Knobelsdorf
JIST-first: Development and

Implementation of an Augmented Reality
MOBMU-317 Kexin Wang MOBMU
Thunderstorm Simulation for General
Aviation Weather Theory Training
OSINT for Critical Infrastructure

MOBMU-318 Mert Ecevit MOBMU
Protection – Electricity Sector
Automated Tools for Cloud Security

MOBMU-319 Hamid Ghazizadeh MOBMU
Testing
Sanitization Tool for Compromised Social

MOBMU-320 Media Account after Recovery and Its Hardik Tapara MOBMU
Aftermath
Financial Security Management: A

MOBMU-321 Comprehensive Study of Cryptocurrency Parth Gandhi MOBMU
Forensics
Automatic Deep Neural Network Model
MOBMU-322 Parakh Agarwal MOBMU
Surgery for Embedded Devices
Probabilistic Time Series Forecasting with Maximilian A. F.

MOBMU-323 MOBMU
Transformer Models Bornstädt
AI-based Cybersecurity Management

Klaus Dieter
MOBMU-325 Consulting - A New Disruptive Technology MOBMU
Schwarz
for the Future
Vulnerability Management Using Open- Navaneeth

MOBMU-326 MOBMU
Source Tools Shivananjappa
Contribution of Machine Learning in the

MOBMU-327 Daria Posashkova MOBMU
Field of Renewable Energy
MOBMU-328 OSINT-Based Email Investigation Samrudha Mhatre MOBMU
DeepSync: Affine Transform Recovery via

Dimitris G.
MWSF-329 Convolutional Neural Networks for MWSF
Chachlakis
Watermark Synchronization
Captchas on Darknet Marketplaces:
MWSF-330 York Yannikos MWSF
Overview and Automated Solvers
Novel Watermarking and Scrambling for

MWSF-331 Deepak Poddar MWSF
Convolution Neural Network Weights
One Detector to Rule Them All? On the

Robustness and Generalizability of Raphael Antonius
MWSF-332 MWSF
Current State-of-the-Art Deepfake Frick
Detection Methods
Improving Video Deepfake Detection: A

MWSF-333 DCT-Based Approach with Patch-Level Luca Guarnera MWSF
Analysis
Contribution of Residual Signals to the

MWSF-334 Detection of Face Swapping in Deepfake Paul Tesse MWSF
Videos
Efficient Temporally-Aware Deepfake

MWSF-335 Peter Grönquist MWSF
Detection using H.264 Motion Vectors
Analyzing Quantitative Detectors for Edgar

MWSF-336 MWSF
Content-Adaptive Steganography Kaziakhmedov
Secure Payload Scaling for Source

MWSF-337 Eli Dworetzky MWSF
Adaptive Payload Allocation
Prompting Steganography: A New

MWSF-338 Hanzhou Wu MWSF
Paradigm
Prediction of Flipped Bits in Robust Image

MWSF-339 Niklas Bunzel MWSF
Hashes by Machine Learning
MWSF-340 Efficient Hash Lookup for PhotoDNA Martin Steinebach MWSF

A Novel Dataset for Non-Destructive
MWSF-341 Eleonora Breci MWSF
Inspection of Handwritten Documents
DeepMammo: Deep Learning Algorithm

Farid Ghareh
MWSF-342 for Digital Mammogram Source MWSF
Mohammadi
Identification
Sovereign Document Verification by Waldemar

MWSF-343 MWSF
Smartphones Berchtold
Optical Aberration Analysis of Light Field

SDA-344 Displays: A Calibration Approach for Qi Zhang SDA
Enhanced Performance
Vergence-accommodation Conflict:
Accommodation-enabled Versus
SDA-345 Erdem Sahin SDA
Accommodation-invariant Near-eye
Displays
4k Stereoscopic Visualization of Robotic

Microsurgery in Vascularized Free Tissue
SDA-346 Justus Ilgner SDA
Transfer for Head and Neck
Reconstruction
Manipulating Viewing Distance and

SDA-347 Camera Toe-in in a Stereoscopic Remote Eleanor O'Keefe SDA
Vision System
KEYNOTE: The Waking Dream: Next Steps

SDA-348 Aaron Parry SDA
Toward Immersion Entertainment
The Effect of Stereoscopic Depth

SDA-349 Distortion on the Near Oculomotor Eric Seemiller SDA
Response
Automatic Stereograph Feature

SDA-350 Nicholas Routhier SDA
Detection
Successful One Man, Low Cost, 3D CGI

SDA-351 Enrique Criado SDA
Feature Film Conversion
What do we know about Apple's new
SDA-352 Andrew Woods SDA
Spatial Video format?
Comparative Analysis of User Experience

in Virtual Reality (VR) and Mixed Reality
SDA-353 Yoshihiro Banchi SDA
(MR) Systems Using Eye-Tracking
Measurements
Virtual Reality (VR) Space for Radiation

SDA-354 Oncology Education and Collaborative Kulbir Singh Sandhu SDA
Learning
VR Visualization of Avian-Human
Perspectives for Co-Design Practices
SDA-355 Tori Simpson SDA
supporting Empathy Building and
Perspective Shifts
SDA-366 Can Aliens See in 3D? Andrew Woods SDA
VDA KEYNOTE: Toward the Use of

VDA-356 Immersive Technologies for Interactive Thomas Wischgoll VDA
Visualization
JIST-first: Kiviat Defense: An Empirical

Evaluation of Visual Encoding
VDA-357 Mirko Mantovani VDA
Effectiveness in Multivariate Data
Similarity Detection
One Model that Fits Them All:

VDA-358 Psychometrics with Generalized Linear Wangqian Ju VDA
Mixed Effects Models
Creating Visual Persona Profiles in
VDA-359 Jeong-Eun Choi VDA
Telegram using NLP
Evaluating the Recommendations of

VDA-360 LLMs to Teach a Visualization Technique Alark Joshi VDA
Using Bloom's Taxonomy
Interactive Media Planning Data

VDA-361 Marina Ljubojevic VDA
Visualization
RAIV: Researchable Archives for

VDA-362 Hunter Price VDA
Interactive Visualizations
Interactive High-dimensional Audiovisual

VDA-363 Feature Analysis of Live Song Concert Alexandra Diehl VDA
Videos
Data Analysis and Visualization of Crime

VDA-364 Sharad Sharma VDA
Data
SESSION Day Date
3D/4D Scanning and Monday 1/22/2024

Motion Estimation I

Motion Estimation I

Motion Estimation I

Motion Estimation I

Motion Estimation II

Motion Estimation II
Quality Measurement of Monday 1/22/2024

Automotive Sensors

Automotive Sensors
Automotive Sensors

Automotive Sensors
Multi-Modal Imaging and

Neural Methods for Safe Tuesday 1/23/2024
Driving

Driving

Driving
Leveraging Bird's Eye

View Imagery for Safe Tuesday 1/23/2024
Driving

Driving

Driving
Multi-Modal Imaging and Tuesday 1/23/2024

Generative Plug-and- Monday 1/22/2024
Play

Play

Play

Play
Generative Artificial
Intelligence for Remote Monday 1/22/2024
Sensing
Sensing
Sensing
Sensing
Sensing
Intersection of
Computational Imaging Monday 1/22/2024
and Materials Science I
Intersection of
Intersection of
Intersection of
Intersection of
Computational Imaging Tuesday 1/23/2024
and Materials Science II
Intersection of
Intersection of
Intersection of
Intersection of
and Materials Science III
Intersection of
Additive Manufacturing:
In- and Ex-Situ Imaging
for Monitoring and Non- Tuesday 1/23/2024
Destructive
Characterization I
Destructive
Characterization I
Destructive
Characterization II
Destructive
Characterization I
Destructive
Characterization I
Destructive
Characterization I
Methods in Wednesday 1/24/2024

Computational Imaging


for Monitoring and Non- Wednesday 1/24/2024
Destructive
Characterization II
Destructive
Characterization II
Destructive
Characterization II
Destructive
Characterization II
Imaging at the Edge Wednesday 1/24/2024
Implicit Neural
Representations for Thursday 1/25/2024
Inverse Imaging
Implicit Neural
Inverse Imaging
Implicit Neural
Inverse Imaging
Implicit Neural
Inverse Imaging
Implicit Neural
Inverse Imaging

Poster Session

Poster Session
Destructive
Characterization II
Additive Manufacturing: In- andWednesday 1/24/2024
Color Imaging I Wednesday 1/24/2024

Color Imaging II Wednesday 1/24/2024
Color Imaging III Wednesday 1/24/2024
Computer Vision and Tuesday 1/23/2024

Image Analysis of Art




Virtual Reality II Wednesday 1/24/2024
High Performance
Computing for Imaging Tuesday 1/23/2024
Keynote Talk I
High Performance Tuesday 1/23/2024

Pathology Imaging

Pathology Imaging
Pathology Imaging

Image Reconstruction

Image Reconstruction
Special Session on High Tuesday 1/23/2024

Performance Imaging

Performance Imaging

Performance Imaging
High Performance
Computing for Imaging Wednesday 1/24/2024
Keynote Talk II
High Performance
Convolutional Neural Wednesday 1/24/2024
Network
High Performance
Convolutional Neural Wednesday 1/24/2024
Network
High Performance
Image Transfer and Wednesday 1/24/2024
Preprocessing
High Performance
Preprocessing
High Performance
Preprocessing
High Performance Vision Wednesday 1/24/2024

Transformer
High Performance Vision Wednesday 1/24/2024

Transformer
Art: Perception and Monday 1/22/2024

Cognition I

Cognition I
Cognition II

Cognition II

Cognition II

Cognition II
Perceptual Experiments Tuesday 1/23/2024

and Models

and Models

and Models
HDR: Imaging and Tuesday 1/23/2024

Perceptual Modeling
Perceptual Modeling

Perceptual Modeling
Perception of AR/VR/MR Tuesday 1/23/2024
Perceptual Video Quality Wednesday 1/24/2024

Visual Quality Across

Displays and Viewing Wednesday 1/24/2024
Conditions I

Conditions I
Conditions I

Conditions I

Conditions II

Conditions II
Visual Perception of Thursday 1/25/2024

Quality

Quality

Quality
Vision and Imaging Thursday 1/25/2024

Human Vision and

Electronic Imaging Tuesday 1/23/2024
Poster Session
Human Vision and

Poster Session
Human Vision and

Poster Session
Human Vision and

Poster Session
Imaging Tuesday 1/23/2024

Imaging and Multimedia

Analytics at the Edge Tuesday 1/23/2024
Poster Session

Poster Session

Poster Session
Image Denoising and Tuesday 1/23/2024

Restoration

Restoration

Restoration
Machine Learning for Tuesday 1/23/2024

Image Processing

Image Processing
Image Formation Tuesday 1/23/2024
Image Processing
Algorithms and Systems Tuesday 1/23/2024
Poster Session
Image Processing
Poster Session
Image Processing
Poster Session
Image Processing
Poster Session
Image Quality in Monday 1/22/2024

Machine Vision
Image Quality in Monday 1/22/2024

Machine Vision

Automotive Sensors
Camera Quality Tuesday 1/23/2024
Assessment I

Assessment I

Assessment I

Assessment I

Assessment II

Assessment II

Assessment II

Conditions II
Conditions II
System Performance Thursday 1/25/2024
High Dynamic Range Thursday 1/25/2024

Evaluation

Evaluation

Evaluation
Vision Systems for Wednesday 1/24/2024

Industrial Applications



Image Sensor Design Wednesday 1/24/2024
and Analysis

and Analysis

and Analysis

and Analysis
Applications and Wednesday 1/24/2024

Processing

Processing

Processing
Quanta, Event, and Wednesday 1/24/2024

Time of Flight Sensors


Imaging Sensors and Tuesday 1/23/2024

Systems Poster Session
Machine Learning for Monday 1/22/2024

Scientific Imaging I





Scientific Imaging II



Intersection of
Intersection of
Intersection of

Scientific Imaging III




Machine Learning for

Scientific Imaging Poster Tuesday 1/23/2024
Session
Multimedia Applications Tuesday 1/23/2024

Multimedia Security Wednesday 1/24/2024

Machine Learning and AI Wednesday 1/24/2024
Machine Learning and AI Wednesday 1/24/2024
Mobile Devices and

Multimedia: Enabling
Technologies, Tuesday 1/23/2024
Algorithms, and
Applications Posters
Mobile Devices and

Algorithms, and
Mobile Devices and

Algorithms, and
Mobile Devices and

Algorithms, and
Watermarking Monday 1/22/2024

Deepfake Detection Monday 1/22/2024
Steganography & Tuesday 1/23/2024

Steganalysis

Steganalysis

Steganalysis
Robust Hash Tuesday 1/23/2024
Robust Hash Tuesday 1/23/2024

Verification & Tuesday 1/23/2024
Identification

Identification

Identification
Stereoscopic Displays Monday 1/22/2024
Stereoscopic Displays Monday 1/22/2024
Stereoscopic Camera Monday 1/22/2024

Systems
Stereoscopic Camera Monday 1/22/2024

Systems
SD&A Keynote Monday 1/22/2024
Human Factors and 3D Wednesday 1/24/2024

Content Solutions

Content Solutions

Content Solutions
Content Solutions
Virtual Reality I Wednesday 1/24/2024
Virtual Reality I Wednesday 1/24/2024
Stereoscopic Displays Wednesday 1/22/2024
Visualization and Data

Analysis Keynote & Wednesday 1/24/2024
Multivariate Data
Visualization and Data

Analysis Keynote & Wednesday 1/24/2024
Multivariate Data
Machine Learning Wednesday 1/24/2024

Information Wednesday 1/24/2024

Visualization
Information Wednesday 1/24/2024

Visualization
Video Analysis Poster Tuesday 1/23/2024

Session
Video Analysis Poster Tuesday 1/23/2024

Session
Abstract: For translationally moving objects with fixed cameras, such as robots and cars, blurring can often be more pronounc
together with an efficient iterative optimization algorithm. Simulation and experiments demonstrate that for scenes with suffi
Abstract: This paper proposes a VGA-size light-tracing time-of-flight (ToF) CMOS image sensor that combines the hybrid ToF m
multi-tap one drain (GD) pixels and non-overlapping multi-subframe technique to implement multiple time windows for the sh
transfer rate allowing all the 12 zones to be scanned and read out at an acceptable frame rate. A porotype chip is fabricated u
speed is higher than 10fps. In addition, outdoor depth image under the high ambient light conditions have been obtained.
Abstract: Solid-state lidar cameras produce 3D images, useful in applications such as robotics and self-driving vehicles. Howev
data from the total laser and ambient light collected in each pixel, to extract additional depth information and boost ranging p
depth range by 2× in an indoor environment and almost 10× outdoors. Depth information is also extracted from features in th
lidar cameras, the principle of extracting depth data from lidar camera intensity images could also be extended to standalone
Abstract: In recent years, several deep learning-based architectures have been proposed to compress Light Field (LF) images a
learning video compression network, adapted and optimized specifically for LF image compression. We enhance this network
manage to address challenges presented by the unique features of LF images, such as high resolution and large data sizes. Our
shows improved performance in terms of Peak Signal-to-Noise Ratio (PSNR) and Mean Structural Similarity Index Measure (M
advancements in this field.
Abstract: Neural Radiance Fields (NeRF) have garnered particular attention due to their exceptional capability in virtual view g
methodology for training NeRF leveraging external depth information. The approach entails generating new virtual views at di
output quality when employing the generated views in comparison to a scenario where they are omitted.
Abstract: Recent advancements in 3D data capture have enabled real-time acquisition of high-resolution 3D range data, even
data problem thus far are image-based encoding schemes that leverage modern image and video codecs. To our knowledge, n
(N-DEPTH) leverages deep learning to efficiently encode depth maps into 24- bit RGB representations that minimize end-to-en
end on the FlyingThings3D dataset with randomized JPEG qualities. On a Microsoft Azure Kinect depth video recording, N-DEP
of image qualities. Our method offers an efficient solution for emerging 3D telepresence and other 3D streaming applications,
Abstract: High dynamic range (HDR) image sensors continue to evolve in range capability and ubiquitousness, particularly with
image quality attribute for HDR image sensors is the quantification of noise, both temporal and fixed pattern. The increased, a
sensor noise were not written with current HDR image sensors in mind, quantification is generally limited to portions of the op
exposures needed, and comprehensive analysis. This paper will show applications of the IEEE P2020 noise standard with expa
Abstract: Noise Equivalent Quanta (NEQ) is an objective Fourier metric which evaluates the performance of an imaging system
the modulation transfer function (MTF) of the system by a power normalized noise power spectrum (NPS) and is a fundament
The dead leaves measurement technique is able to provide an estimate of the MTF and NPS of an imaging system using a sing
of a single measurement position. This is particularly useful if measurements are made post-ISP and non-linear processing ren
This work demonstrates the use of the dead leaves technique to measure NEQ, firstly through simulation of an imaging system
practical method to estimate NEQ, ranking imaging systems performance both pre- and post-ISP.
Abstract: The Modulation Transfer Function (MTF) is an important image quality metric typically used in the automotive doma
wide field-of-view (FOV) cameras have become increasingly popular, particularly for low-speed vehicle automation application
Abstract: In recent years, road vehicles are equipped with an increasing number of cameras, for driver assistance, driver moni
count. Moreover, auto parts are subject to dramatic variations in temperatures, and the thermal expansion of the imaging har
integrated and fully automated system based on collimator, which makes feasible the testing of a wide variety of automotive
to results of high accuracy and repeatability, which is consistent with tests conducted with conventional methods. Preliminary
Abstract: Multi-modal pedestrian detection has been developed actively in the research field for the past few years. Multi-mo
can simultaneously use complementary information from visible and thermal frames. However, many existing multi-modal pe
paper proposes a multi-modal pedestrian detection network for a one-stage detector enhanced by a dual regressor and a new
experiments demonstrate that the proposed method outperforms current state-of-the-art methods on artificial data with larg
Abstract: A naturalistic driving study includes a blurred cabin image to find non-consenting passengers in the vehicle. We inve
detection method. We also compared the image classification using convolutional neural network and vision transformer bac
somewhat niche. We perform some analysis to support our conclusion.
Abstract: With AI becoming the defacto approach to solving a myriad of problems across industrial, automotive, medical, defe
tasks optimally also include making them fault-tolerant. As embedded devices might be deployed under varying temperatures
level of ASIL-B implements typical ECC circuitry in memory to handle single-bit-error correction and double-bit error detection
which makes the solution an expensive proposition. Some require active cooling as well due to increased redundant computin
fault tolerant system without linearly doubling or tripling compute area cost.
Abstract: Automated extraction of intersection topologies from aerial and street-level images is relevant for Smart City traffic-
The intersection topology is expressed in the amount of approach lanes, the crossing (conflict) area, and the availability of pai
Segmentation of road surface and other basic information can be obtained with 80% score or higher, but the segmentation an
This paper addresses this complicated problem by proposing a dualistic channel model featuring direct segmentation and invo
These channels are developing specific features such as drive lines and lane information based on painted striping, which are fi
The algorithms and models are evaluated with two datasets, a large mixture of highway and urban intersections and a smaller
Experiments with measuring the GEO metric show that the proposed late-fusion system increases the recall score with 4-7 pe
This recall gain is consistent for using either aerial imagery or a mixture of aerial and street-level orthographic image data.
The obtained recall for intersections is much lower than for highway data because of the complexity, occlusions by trees and t
Future work should aim at consolidating this model improvement at a higher recall level with more annotated data on interse
Abstract: Predicting the trajectory of an ego vehicle is a critical component of autonomous driving systems. Current state-of-th
struggle with perspective issues affecting object features in the scene. To address this, we advocate for the use of Bird's Eye V
encoding to represent objects in a BEV, achieving competitive performance compared to traditional DNN-based methods. Wh
between the objects in a scene are captured effectively
Abstract: Traffic simulation is a critical tool used by psychologists and engineers to study road behavior and improve safety sta
approach to utilize fundamental image processing techniques to identify key features of an environment from a top-down vie
automatically generated while buildings and trees are selected from a library based on detected attributes. This research wou
Abstract: Dense 3D reconstruction has many applications in automated driving including automated annotation validation
whereas camera provides dense depth but noisy particularly at long ranges. In this paper, we harness the strengths of both se
representation based on signed distance field, which can be further rendered into RGB images, and depth maps. A mesh can b
qualitative and quantitative results on challenging automotive scenes.
Abstract: Inversion by Direct Iteration (InDI) is a new supervised image restoration method that avoids the regression to the m
ill-posed problem, meaning that there are multiple possible high-quality images that could be reconstructed from a given low-
gradually moving towards a more realistic and detailed reconstruction. InDI is a flexible and powerful framework for a wide ra
Abstract: Estimating high-quality images while also quantifying their uncertainty are two desired features in an image reconstr
solutions to a general inverse problem. PMC is able to incorporate expressive score-based generative priors for high-quality im
the traditional plug-and-play priors (PnP) and regularization by denoising (RED) algorithms. We also establish a theoretical ana
likelihoods and imperfect score networks. We demonstrate the performance of the PMC algorithms on multiple representativ
quantification.
Abstract: Prognosis for melanoma patients is traditionally determined with a tumor depth measurement called Breslow thickn
invasive melanoma in whole-slide images. First, we design a custom segmentation model from a transformer pretrained on br
superior results compared to previous approaches and qualitatively better results as verified through a dermatologist.
Abstract: Plug-and-Play (PnP) is a modular image reconstruction framework in which an image denoiser can be used as a prior
Here we introduce Generative Plug-and-Play (GPnP), a generalization of PnP to sample from the posterior distribution. GPnP h
from the posterior.
Abstract: Neural radiance fields and scene representation networks offer unprecedented capabilities for photorealistic scene r
generalization strategies that allow us to generate photorealistic multi-view-consistent humans or cats using state-of-the-art 3
Abstract: Computational imaging is permeating cameras and microscopes across many scientific applications; but how can we
imperfect optical systems or approximate forward models. Traditional model-based image reconstruction algorithms work tog
Abstract: Single image super-resolution (SISR) research has witnessed advancements with the development of deep learning.
deep learning based super-resolved reconstructions. Alternatively, by using active coherent or partially coherent illumination,
exceed that of a single-aperture system; a typical single-aperture system only being physically capable of measuring a limited
measured by each array camera’s multi-view sampling of the Fourier domain, while utilizing our prior work with neural repres
illumination scheme, array camera parameters, generative AI architectures and strategies a flexible and intelligent imaging sys
Abstract: I will discuss diffractive optical networks designed by deep learning to all-optically implement various complex functi
computational camera and microscope designs and (3) inverse design of optical systems that are task-specific. In this talk, I wi
biomedical imaging and sensing.
Abstract: In this talk, we discuss the approach of combining the physical encoding in imaging (deterministic transformation) an
compressive photography, ultrafast optics and low photon count imaging. In addition, we explore the possibility of using this f
Abstract: Materials Discovery involves identifying candidate materials providing properties exceeding those conventionally ava
processing conditions and microstructure. This would provide the stability and predictability necessary for design iterations as
ill-posed. This presentation gives, by way of introduction, our initial results for topological properties of materials data, as well
Abstract: We hypothesize that microstructure images lie on a low-dimensional manifold, i.e. the manifold hypothesis. For a m
used MLE estimates to find the dimensionality. We find a modification to the distance metric is needed for sparse datasets, bu
Abstract: Quantitative microstructure metrics are an integral component of understanding processing-structure-property rela
continuous manifold from microstructure descriptors that can be mapped back to processing conditions, a bi-directional relati
local changes have a small or large impact on resulting microstructure, as well as help define safe boundaries for potential pro
Abstract: This paper presents a novel method to learn dissipative systems using a Lagrangian neural network. A dissipative sys
physical system and preserves energy or matter. The method can handle both Newtonian and diffusing systems and can navig
Abstract: Gaussian processes have shown to be powerful tools for stochastic function approximation and autonomous control
performance in real-life applications. The reason, however, is often not the method itself but missing flexibility and domain aw
Abstract: In recent years, there has been a rising interest in utilizing Machine Learning algorithms for Additive Manufacturing p
immediately applicable for such problems due to the complex physics involved in printing metals. This in turn breaks standard
computationally tractable manner.
Abstract: In the field of polymers, 2D images are often used to discern information about the microstructure of bulk polymer m
techniques that evaluate thin film (i.e., electron imaging of brush particle monolayers) and bulk materials (electron tomograph
systematic discrepancies of characteristic spacings between nanoparticles in thin film and bulk materials. An approach to ratio
Abstract: Binder jet additive manufacturing can achieve complex geometries with metal and ceramic materials, but low green
behaviors and correcting them within the same build is necessary to build with new materials. In this study, Dragonfly image p
316 stainless steel and alumina and segmented using a U-Net convolutional neural network to identify varying spreading anom
builds to reduce labeling and training time investments. Trends in the number of pixels classified as each label reflected chang
be corrected, allowing for improved part quality and the development of closed-loop additive manufacturing.
Abstract: Despite its widespread use, secondary electron imaging remains a qualitative nanoscale imaging technique due to sh
We describe a model for the response of the secondary electron detector and show how detector parameters can be measure
nanoscale material characterization at low doses suitable for beam-sensitive biological samples.
Abstract: Ultrasound Computed Tomography (USCT) offers a radiation-free approach with profound potential for high-resoluti
problem, known as Full Waveform Inversion (FWI) to reconstruct tissue properties. However, the time-intensive nature of wav
simulations and address the frequency-domain USCT image reconstruction challenges. Experimental validations on phantom d
Abstract: Non-destructive characterization and evaluation (NDC/NDE) using X-ray computed tomography (XCT) is crucial in me
to high density of the metallic materials and complexity of printed components have limited XCT’s role in high-throughput larg
In this work presents an overview of fast, deep learning-based algorithms developed to enable XCT for scalable high-throughp
deep learning networks, including generative adversarial networks (GANs). This combination enhances the quality of the recon
based flaw detection and segmentation.
Our algorithms have shown considerable improvements in NDE for metal AM, enabling an order of magnitude faster scans com
burdens. We present results demonstrating how these algorithms has allowed for high-throughput characterization of hundre
Abstract: Significant interest in the production of aerospace parts by laser powder bed fusion (LPBF) has led to the modificatio
constituents to the powder bed, leading to exothermic reactions which form inoculant particles and slow the cooling rate. In o
understood.
In this work, x-ray microscopy was used to visualize and quantify the shape, size, and to classify the starting powders used in t
RAM particles and pores were correlated to local grain structure and mechanical behavior. These observations lead to a novel
Abstract: The complexity of metals additive manufacturing (AM) processing remains a significant barrier for fully understandin
causing similarly local fluctuations in resulting material structure. Monitoring the process in situ offers a window into characte
the wide variety of sensor modalities and data representations, as well as the problem of cross-modality registration. We show
source framework for correlative analysis. We demonstrate this framework on data collected in laser powder bed fusion Ti-6A
note potential limitations and areas of improvement for the overall framework.
Abstract: Additive manufacturing processes have gained traction in many industries, including aerospace, medical and dental
manufacturing technologies. The layer-by-layer nature of additive processes, however, offers unique opportunities for in situ i
introduces a new class of AM sensing modalities, which we call temporally integrated images, that are engineered to capture p
video streams generated from a combination of visible and near infrared cameras. These image generation algorithms can fac
Abstract: The conformity, integrity and cleanliness, of additively manufactured (AM) parts need to be demonstrated before th
destructive testing (NDT) methods are required to inspect both external and internal features. Depending on their size and ph
twins, which allows to better understand the origin of AM deviations, to predict them, and to make appropriate adjustments t
computed tomography (CT) can resolve the three-dimensional position of features. Finally, when the thickness of the parts pr
This talk will address X-ray radiography, CT and tomosynthesis used for quality inspection of AM polymer, ceramic and metal p
Abstract: Non-destructive characterization of an additively manufactured part is performed using X-ray computed tomography
metal components. The acquisition process can be accelerated by reconstructing the 3D volume from sparse projections, but s
State-of-the-art image algorithms are either computationally expensive or are unable to remove artifacts at higher sparsity. D
We propose model-based deep learning (MBDL) algorithm that solves a regularized image recovery problem. It uses a convolu
enhances segmentation quality, thus facilitating more accurate detection of defects.
Abstract: Traditional super-resolution processing computes sub-pixel alignment over a sequence of image captures to allow sa
fractions of a pixel between exposures. The implicitly perfect alignment of pixel-shift images reduces postprocessing to interle
difficult, so the synthesized super-resolution image is instead built from already-interpolated image data, with a reduction in t
The current work instead directly models the certainty with which pixel values are known. Sub-pixel alignment may be compu
motion, even raw pixel values can have variable value certainty. Thus, a certainty metric is calculated for each raw pixel value
Abstract: Acquisitions of mass-per-charge (m/z) spectrometry data from tissue samples, at high spatial resolutions, using Mass
Squares (SLADS-LS) algorithms follow compressed sensing principles to minimize the number of physical measurements perfo
learning model, to provide the most relevant information to an intended reconstruction process. Preliminary results for DLADS
Ionization (nano-DESI) MSI. A new multimodal DLADS variant incorporates optical imaging for a 5% improvement in final recon
m/z channels targeted during scans, with negligible impact on location selection and reconstruction quality.
Abstract: The dynamic range of the intensity of LWIR cameras are often more than 8bit and its images have to be visualized us
visualization method based on gradient-domain processing (Gradient mapping). Processing based on intensity and gradient po
Abstract: Process monitoring in laser powder bed fusion (LPBF) has been available for almost a decade, yet the technology's ap
to explore alternative and more beneficial applications, such as process development and closed-loop control. This technical p
of process monitoring in a closed-loop control methodology is elucidated. The proposed solution employs real-time intelligent
process, with the acquired data facilitating real-time adjustments to laser power. The outcome is the production of componen
remarkable 2x-5x reduction in build times. The presentation aims to provide insights into the evolving landscape of LPBF techn
Abstract: Synthesis of novel materials systems requires the optimization of key characteristics across various physical length sc
models on image synthesis tasks has revealed opportunities for materials scientists to leverage these powerful new tools. This
microstructure generation reveals that the model is context aware of morphological descriptors which may be encoded in the
Abstract: Most in-process techniques for melt pool monitoring are only sensitive to the melt pool portion directly visible on to
imaging at synchrotron sources can be alternatively used to capture the subsurface melt pool behavior; however, such metho
Recognizing the substantial gap between top level melt pool monitoring techniques and high-speed X-ray imaging at synchrot
wavelengths on the order of the melt pool dimensions are used to interrogate the melt pool from beneath the surface. Ultraso
describes the experimental configuration of multiple ultrasonic sensors used for melt pool monitoring in addition to results fro
Abstract: In metal additive manufacturing (AM), it is common to use destructive microscopy techniques such as serial-sectioni
parts. While these approaches allow for micron-scale characterization of features and defects/flaws, they are destructive, cost
dual-energy X-ray computed tomography (XCT) that not only allows for classification of defects (pores, cracks, and inclusions),
We XCT scanned a crept Aluminum alloy AM part at two energies. Both projection data were reconstructed using MBIR. Our a
energy reconstruction allowed for better delineation of MBPs and MBRs. We further analyzed the segmented images for quan
Abstract: At the pinnacle of computational imaging is the co-optimization of camera and algorithm. This, however, is not the o
introduces the concept of computational image formation (CIF). Compared to the standard inverse problems where the goal is
result.
Abstract: We present a novel architecture highly efficient at complex 2D data computations - one of the drawbacks of the high
present bottleneck in today's computational imaging algorithms. Our architecture uses programmable interconnects that reco
the path thus significantly reducing the number of clock transition required in a data path propagation from source to destina
Abstract: In this talk I will discuss the interactions between optical system point spread functions (PSFs) and modern nonlinear
speed. I will also discuss the use of dynamic PSFs for privacy-preserving computer vision, leading to systems which can make v
Abstract: Energy-resolved neutron imaging enables measurements of transmission spectra in each pixel of imaging dataset. Th
a challenging computational task mainly due to the relatively poor quality of measured spectra, in particular low neutron stati
alloy sealed within a brass container. Such analysis can be beneficial for non-destructive testing of various solid materials and
Abstract: Biological vision systems are efficient, robust, and well-adapted to specific tasks in specific environments. This makes
zebrafish, we will discuss the benefits and pitfalls of drawing on animal models in computational imaging. Strategies for findin
Abstract: Will add later.
Abstract: Dynamic contrast-enhanced multispectral optoacoustic tomography (DCE-MSOT) is an emerging imaging modality fo
reconstructed image can then be viewed as a function of space, wavelength, and time. Reconstructing for such multidimensio
due to correlations across the spatial, temporal, and spectral dimensions. Previous works have demonstrated that neural field
multispectral dynamic image reconstruction. The proposed approach is demonstrated with two numerical DCE-MSOT imaging
small animal model. The second study conducts a virtual DCE-MSOT imaging experiment to demonstrate the feasibility and ac
Abstract: Synthetic aperture sonar (SAS) measures a scene from multiple views in order to increase the resolution of reconstru
due to a limited number of measurements and bandlimited hardware, which limits the capabilities of existing reconstruction m
optimization enables us to incorporate physics-based constraints and scene priors into the image formation process. We valid
reconstructions than existing approaches.
Abstract: Implicit neural representation (INR) has recently demonstrated excellent performance in computational imaging. In t
model the imaging object in the spatial domain, while in the latter, INR is used to represent the measurement field. These two
Abstract: Motion during computed tomography data acquisition leads to artifacts and when imaging the heart, this motion is n
Algorithms which measure and/or estimate the displacement have had limited success due primarily to challenge of estimatin
We have developed an analysis-by-synthesis approach which progressively regresses a solution consistent with the acquired s
networks.
As a result, optimization can be performed under spatial and temporal smoothness constraints without the need for explicit m
Abstract: X-ray phase-contrast tomography (XPCT) is widely used for high contrast 3D imaging using either synchrotron or labo
dominant approaches to 3D reconstruction using XPCT relies on the use of phase-retrieval algorithms that make one or more l
blur, artifacts, or other quantitative inaccuracies. Our solution to this problem is to formulate new iterative non-linear phase-r
quantitatively accurate
reconstruction with reduced artifacts. Unlike existing NLPR algorithms, our approaches avoid the laborious manual tuning of r
space and solve the phase-retrieval problem. These constraints are easily user-configurable since they follow directly from the
Abstract: We propose a method to quantify each fluorescence using a liquid crystal variable retarder as a spectral modulator,
quantify the fluorescence at each pixel using intensities of various spectral transmittances as a feature vector, and applying sp
feasibility through simulation experiments.
Abstract: Domain adaptation, which transfers an existing system with teacher labels (source domain) to another system witho
the target domain that were not present in the source domain. In conventional methods, unknown labels have been treated a
there is a difference in distance from the known labels in the source domain depending on the type of unknown label. Therefo
classified. Pseudo-labels are given to the target samples based on the classification results, and the accuracy of domain adapta
Abstract: The additive manufacture process of building a part introduces the processing parameters characteristic defects tha
relations, optimization of processing parameters and fit-for-purpose materials development. Type and physical origin of these
In this work, we explore object detection and semantic image segmentation methods to find location and shapes of the defec
statistically equivalent microstructures images and improve the accuracy of the classification of the defects per their kind. We
results of developed generative models yield microstructure images that are to the high degree indistinguishable from the ima
Abstract: The additive manufacture process of building a part introduces the processing parameters characteristic defects tha
relations, optimization of processing parameters and fit-for-purpose materials development. Type and physical origin of these
In this work, we explore object detection and semantic image segmentation methods to find location and shapes of the defec
statistically equivalent microstructures images and improve the accuracy of the classification of the defects per their kind. We
results of developed generative models yield microstructure images that are to the high degree indistinguishable from the ima
Abstract: In digital cameras, the wavelength dependency of image generation caused by hardware directly affects the spatial r
spectroradiometer were used to reveal the wavelength dependency of the spatial resolution characteristics. We then confirm
minimizing the color change caused by the adjustment.
Abstract: Pixel binning enables high speed, low power readout in low resolution modes, and more importantly, a reduction of
high resolution, CMOS image sensors for mobile phones have moved beyond the once-binnable Quad Bayer and RGBW-Kodak
Hexadeca Bayer pattern featuring 4x4 tiles of like colored pixels.
In this paper we propose double binnable variants of the recently proposed LMS CFA and also of the older RGBW CFAs. These
the proposed LMS and RGBW CFAs provide higher SNR than existing binnable CFAs. We present a frequency domain analysis o
the industry and arrive at the counter-intuitive result that Nona (3x3) and Hexadeca (4x4) Bayer can be superior to Quad (2x2)
Abstract: Digital cameras have become indispensable tools in today's society, playing pivotal roles across scientific research, in
interpretation. Such deviations not only impact aesthetic quality but also significantly affect the accuracy of data-dependent a
device-dependent RGB color space to a device-independent XYZ color space that models the human visual system, thereby br
nature of the transformation.
Abstract: Accurate color reproduction on a display requires an inverse display model, mapping colorimetric values (e.g. CIE XY
that, unlike traditional LCD displays, multi-primary OLED display color responses are non-additive and non-linear for some colo
validated on newly measured (unseen) test colors. We found that the performances of several variations of 4th-degree polyno
Abstract: Screen printing is an established method widely used for printing with media or substrates not suitable for inkjet prin
effects, the printing procedure needs to be adjusted and the knowledge of angularly dependent color properties of such prints
characterisation after acquiring its inline Bidirectional Reflectance Distribution Function.
Abstract: Variability in the encoding of unique hues within the color vision system results in significant diversity in hue appeara
default NCS unique hue in a modern color appearance model with individualized selections. By customizing the color represen
perception research and have practical implications in various domains such as, graphic design, and visual color reproduction.
Abstract: A method is proposed to estimate surface-spectral reflectance from camera responses using a local optimal reflectan
database is utilized to locally determine the candidates to optimally estimate the spectral reflectance. The proposed method
Experimental results show that the estimation accuracy has improved compared with the previous results.
Abstract: The CIE TC8-16 (International Commission on Illumination Technical Committee) proposed the definition of Consiste
one embodiment is consistent with the relationship between the corresponding set of colors in another image [1]. We conduc
simulating the corresponding CMYK printing conditions were used to create the images [2], which were presented to observer
Abstract: We studied White/Black/Gray segments in Assimilation and Contrast illusions. Many White pixels generate high amo
White/Black angular separation. Receiving pixels fall much higher on CIE GlareSpread Function substantially increasing glare o
extreme outliers, and are not helpful in understanding how human spatial-image-processing works.
Abstract: Since the introduction of the Retinex theory by Land and McCann in 1971, a multitude of different families, versions,
the local context, and (ii) how they recompute the pixel values based on this context. STRESS (spatio-temporal Retinex-inspire
sampling technique, resulting in two envelope images - one maximum and one minimum envelope, completely enclosing the i
noise, with an adapted version of constrained linear anisotropic diffusion for computing the envelopes,resulting in almost nois
Abstract: Methods for estimating spectral reflectance from XYZ colorimetry were evaluated using a range of different types of
and metameric differences under different illuminants are the lowest. The study demonstrates that, a training data can be ma
spectral and colorimetric outcomes. The findings suggest that using polynomial bases or colorimetric weighted bases with leas
tristimulus values were assessed with tristimulus calculated using the measured reflectances and a destination illuminant. The
with specific training dataset can serve as a good alternative to predict XYZ under different illuminants with reduced metamer
such that it is simple to implement, fast in processing, spectrally accurate with low metameric mismatch.
Abstract: We extend our prior computational image-analysis methods for estimating head pose angles to also include gender i
demonstrate our system on 23,121 portraits in the Western canon. We find a small gender difference in yaw, corroborating p
and may include additional visual attributes.
Abstract: Artificial Intelligence is increasingly being consumed in artistic and design practices. However, our understanding of A
models - the controllable aspects of the world they create - becomes essential.
The contribution of this work is twofold: first, we utilize state-of-the-art disentanglement methods for color editing of textile p
palettes. To the best of our knowledge, no disentanglement experiments have been carried out in the context of color editing
The result is an efficient, training-free, and adaptive color editing tool with a vast range of applications, from intelligent colorw
in the models; on the latent similarity of the colors in this space; and on their combinatorial properties.
Abstract: This study presents a novel approach to understanding pictorial light in art history through computational methods.
lighting information, allowing for a more nuanced capture of complex lighting conditions compared to previous methods limit
viability and accuracy of the method. Although the exploration of visual patterns is a recurring focus in digital art history, this r
Abstract: Can artists be recognized from the way they render certain materials, such as fabric, skin, or hair. In this paper we st
based on Swin Transformer and Cascade Mask R-CNN to address artist recognition task. We report the performance on a data
Abstract: Artwork analysis and authentication are pivotal in the fields of art history and connoisseurship. This study introduces
support vector machine (SVM) for binary decision-making. Additionally, image edge feature analysis algorithms, vital for unde
Raphael paintings. Thus, we show that carefully considered machine learning approaches can be useful for complementing tra
Abstract: This contribution explores the effectiveness of Vision Transformers in the art authentication task. We employ a caref
incorporation of AI-generated forgeries addresses the common challenge of limited access to human-made forgeries. Our find
training enhances the detection of human-made forgeries.
Abstract: During emergencies like fire and smoke or active shooter events, there is a need to address the vulnerability and ass
spatial analysis and visual communication of emergency-related information to the user. This paper presents the developmen
incorporates existing permanent features such as room numbers and signages in the building as markers to display the floor p
current location of the person, the number of exits, and user-specific personalized evacuation routes. The paper also describe
show the effectiveness of our situational awareness-based mobile augmented reality application in multilevel buildings for eva
Abstract: Stereoscopic rendering is a technique used for providing depth cues as used for virtual reality (VR) applications. In sh
applications, thus presenting additional challenges for real-time applications. The critical flicker frequency (CFF) describes the
spatial reprojection of pixels distributed to multiple GPUs. This enables hole-filling and denoising to be parallelized as well whi
loop that adapts the path tracing parameters to the current rendering speed.
Abstract: Aviation Maintenance Technicians (AMTs) play an important role in guaranteeing the safety, reliability, and readines
Currently, aviation maintenance training methods are centered around classroom instruction, printed manuals, videos, and on
research explores the implementation of Virtual Reality (VR) platforms as a method for enhancing the aviation training experie
based, open-source, immersive platform called Mozilla Hubs. While there is a general belief that VR enhances learning in gene
specifically for aircraft maintenance training.
Abstract: Over the past two decades, scientific and engineering imaging applications have undergone a profound evolution in
demands of imaging applications and algorithms. In a parallel development, AI is gaining strides in imaging workflows. As high
scale AI. In this talk we present our vision on how reconcile the tensions between multiple technologies that would act as a fo
Abstract: When addressing the challenges posed by giga-pixel digital pathology in whole-slide imaging, a significant portion of
hindrance often emerges within the input-output (I/O) system. This article outlines our efforts to overcome the data access ch
greatly facilitates efficient data management across WSIs.
Abstract: Multiplex immunofluorescence (MxIF) is an advanced molecular imaging technique that can simultaneously provide
eosin (H&E) stain is typically unavailable with MxIF on the same tissue section. As biological H&E staining is not feasible, previo
highresolution WSI-wise synthesis. The MxIF to H&E synthesis is no exception. Limited by computational resources, the cross-s
back to a WSI. In this work, we propose a deep learning based unpaired high-resolution image synthesis method to obtain virt
segmentation supervision as spatial constraints. Then, we introduce a random walk sliding window shifting strategy during the
cell segmentation task. The proposed inference method reduces the tiling effects by using 50% fewer computation resources w
synthesis applications. The source code with our proposed model is available at https://github.com/MASILab/RandomWalkSlid
Abstract: Analyzing WSI images presents unique opportunities and challenges as these images capture microscopic details and
of complex deep learning models. A commonly used approach involves dividing WSI images into smaller patches and training
resolution images to minimize I/O and memory overhead; 2) Creating a balanced dataset that includes annotated regions for c
patches. We've explored Vision Transformers and convolutional models for feature extraction. We've also utilized self-supervi
through distributed training techniques on the world's fastest supercomputers, Summit and Frontier.
Abstract: Computed Tomography (CT) is a widely-used 3D imaging technology known for its utilization of computationally inte
complexity while enhancing data locality. These groundbreaking algorithms have been transformed into efficient back-projecti
harnesses a high-level OpenCL implementation to generate vectorized code. Additionally, we capitalize on OpenCL's local mem
generations, consistently demonstrate that our optimized back-projection implementation achieves a remarkable speedup, ou
Abstract: Laminography is a specialized 3D imaging technique optimized for examining flat, elongated structures. Laminograph
the preferred computational method for generating high-quality 3D volumes, however these methods are computationally de
performance computing resources such as GPUs. In this work, we investigate the integration of the Unequally Spaced Fast Fou
sampling issues typical in laminography, while the combination of ADMM and CG introduces robust regularization techniques
Compared to the original algorithm, the optimized USFFT method achieved an 1.82X speedup. By harnessing heterogeneous c
of our methods using real-world datasets collected at 32-ID beamline at Advanced Photon Source using Argonne Leadership C
Abstract: Parallel optical and electronic processing is the key to increasing pixels captured per unit volume, time, cost and elec
talk reviews progress on these two components and discusses emerging neural data structures and array data fusion architect
power neural processing may enable truly ubiquitous imaging and machine vision.
Abstract: For twenty years, a small group of us have developed the Image Systems Engineering Toolbox (ISET). This software i
and freely available at GitHub under the MIT License. ISET is used to (a) design and evaluate image systems, (b) develop algori
and (e) train students in image systems engineering.
Abstract: Computational imaging of planetary interiors provides observational constraints for testing hypotheses on the struct
imaging of planetary interiors uses the seismic spectrum (acoustic or shear). Detection and simulation of the seismic wave-fiel
three scenarios where powerful computing systems and algorithms help a seismologist like myself build better images on our
process using stochastic optimization algorithms. And the third is in speeding up the large imaging process by coupling machin
Abstract: Computational wave imaging provides a way to infer otherwise unobservable physical properties of a medium from
issues. Some success has been attained when an abundance of simulations are available. Nevertheless, when applied to a mod
address the critical issues of weak generalizability and data scarcity.
Abstract: We present AutoFastConv, a based on the TVM template auto-generation open-source library that can automatically
convolutional neural networks. Currently, there are two mainstream CPU architectures, x86 and ARM, that cover a wide range
layers of different shapes on different hardware configurations. AutoFastConv addresses this problem by using TVM-based tem
parameter configuration space for the best performance for a given target architecture and problem size. Performance evalua
2.27, and1.01?1.66 times faster than OnnxRuntime, MNN, and TVM, respectively. Layer-by-layer evaluations for popular DL ne
Abstract: In this paper, we propose a novel large kernel metamaterial neural network (LMNN) that maximizes the digital capac
maximize the learning capacity of MNN while modeling the physical restrictions of meta-optic. With the proposed LMNN, the
optimized hybrid design improved classification accuracy while reducing computational latency.
Abstract: Next-generation scientific instruments, such as high-speed imaging detectors in synchrotron radiation facilities, gene
challenging due to the limitations of local compute resources. To address this issue, we present high-performance data transfe
In this talk, we explore data transfer bottlenecks caused by inadequate usage of complex hardware resources, including non-u
available CPU cores, NICs, and bandwidth. Our prototype is tested for high-performance data streaming between (and also wi
streams achieve over 2x speedup compared to standard data transfer techniques. We can also saturate close to 95% of availa
Abstract: The utilization of supercomputers and large clusters for big-data processing has recently gained immense popularity
prominent in the context of large-scale datasets, which can range from hundreds of gigabytes to several terabytes in size. Sim
storage, retrieval, and efficient GPU utilization. In this work, we benchmarked a collection of high-performance general datalo
solutions such as Webdatasets, FFCV, and DALI have demonstrated significantly superior performance when compared to trad
supercomputer environments. Our initial findings indicate that DALI surpasses our baseline PyTorch dataloader by up to 3.4x i
Abstract: Tensor decomposition is widely used in machine learning, recommendation systems, and social networks. Large real
which are difficult to overcome due to the inherent trade-off among the multiple types of computations and communications,
communication and computation for the popular distributed Canonical Polyadic Decomposition (CPD) algorithms. Our algorith
approaches, we break down computation and communication in consecutive stages, and mask the communication costs by a
outperforms the state-of-the-art distributed implementations by up to 1.8X on 1280 processors. Our method can be applied t
Abstract: Vision transformer models trained on long sequences often achieve higher accuracy than short sequences for many
methods for long sequence training offer limited speedup and memory reduction, and may compromise accuracy. This talk pr
communication, resulting in a 5.9x faster and 10.2x more memory-efficient implementation compared to state-of-the-art sequ
GPUs, achieving 161% super-linear parallel efficiency due to improved GPU utilization rate and a throughput of 32 petaflops.
Abstract: In this work, we introduce an efficient distributed sequence parallel approach for training transformer-based deep le
models. These transformer-based deep learning networks for image segmentation comprise of a combination of a vision trans
models must train on larger, high-resolution input data and/or use more model parameters. The amount of memory available
capabilities to train transformer-based image segmentation models on large-scale input data using an efficient distributed seq
training on larger, high-resolution data. We showcase our sequence parallel approach by training image segmentation models
Abstract: Márton Orosz's keynote lecture unfolds a captivating case study, shedding light on an overlooked precursor of media
1960s and early 1970s, Orosz introduces the revolutionary theories of György Kepes, a polymath, and founder of the Center fo
The central question addressed is how to develop an agenda that bridges aesthetics and engineering, not merely as a gap-fillin
alternative for building a sustainable world—a groundbreaking idea in Post-War art history. Kepes's prominence lies in his cyb
culture" in art literature. Moreover, he stands as one of the earliest to employ nanotechnology in creating artwork.
Orosz explores Kepes's insights into visual aesthetics, his concept of the "revision of vision," and the impact of "the power of t
The in-depth analysis ensues into Kepes's quest to establish a universal visual grammar, crafting a novel iconography of scienti
with symbolic meaning. Orosz's paper contemplates Kepes's recently discovered legacy, emphasizing the democratization of v
Abstract: This year, the Modern Art Collection of the venerable Vatican Museums celebrated the 50th anniversary of its found
work typifies Vukadinov's artistic essence, which deeply roots itself in the rich history of art of the region, Thracia—a region w
inventive techniques. The artwork exudes an inexorable sense of transcending both space and time, embodying a profound an
Abstract: In the field of image processing, a common goal is establishing the identity of face images relative to a particular targ
different, which is a much more challenging proposition. In the application to historical works of art, the issue arises in the con
dimensional transformation to match the features. This approach was taken to resolve the question of Leonardo da Vinci’s you
In the case of Michelangelo, there are a substantial number of painted and sculptural portraits of him in later life, providing ad
portraits: his own sardonic self-portrait in the flayed skin held by St Bartholomew, itself a portrait of the scathing satirist Pietro
reproduced face of Adam being brought to life by God, the face of Jesus at the center of the Last Judgment, and the heroic fig
frescos. Taken together with the sixteenth century tendency of artists to paint St Sebastian as their own self-portrait, this resu
renowned ‘David’ and ‘Pieta’ sculptures.
A similar analysis can be extended to identify other plausible portraits of Vittoria Colonna in the same frescos, providing new i
Reference
Tyler CW, Smith WA, Stork DG (2012). In search of Leonardo: computer-based facial image analysis of Renaissance artworks fo
Abstract: x
Abstract: In The Psychology of Perspective and Renaissance Art (1986), I proposed a compensation hypothesis to account for t
Various critics have expressed doubts about this inverse projection approach. In response, I abandon the compensation hypot
plane, is one in which the vanishing point is in the center of the picture. I call this "picture taking", even if the picture is painte
viewer's eye height, which is given by the position of the horizon (which by our assumption is horizontal, and by definition goe
we can distinguish between the real you (R) in the real world, and the virtual you (V), in the virtual world. R is looking at the pi
rule of picture taking — are robust. When artists deviate from picture taking, e.g., by moving the VP away from the center of t
Abstract: In 1922, Rene Magritte glanced at a grayscale reproduction of de Chirico’s 1914 painting "The Song of Love" and, mo
epiphany. Like a neuroscientist, Magritte reveals our visual brain in action: figures become background, only to reverse places
Abstract: The natural ordering of shape is not historically used in visualization applications, but it would be helpful to show wh
entropy allowing us to construct an ordered scale of shape glyphs. Method—we evaluate the visual entropy glyphs in replicate
as predicted by the visual entropy score with large effect size. Conclusion—Visual entropy predicts shape order and provides a
Abstract: Understanding multimodal interaction and its effects on user experience and behavior is becoming increasingly impo
research on how visual and tactile properties of materials and objects interact is still limited.
This study investigates how color appearance of texture images affects observers’ judgements of different tactile attributes. Fo
color space. The observers were asked to rank these texture images on different haptic properties to test the effects of color o
and robotic surgery, training simulators, and others.
Abstract: Contrast sensitivity functions (CSFs), which provide estimations of detection thresholds, have far-reaching applicatio
model. However, current spatiotemporal CSFs are inaccurate when predicting high-frequency limits such as critical flicker freq
both in the fovea and eccentricity. We trained these models using 15 datasets of spatial and temporal contrast sensitivity mea
of CFF at high luminance values. As a result, the prediction errors for CFF obtained from the modified models improved remar
Abstract: Today, many imaging experiences include some form of High Dynamic Range. With the right combination of equipm
content creator. This major improvement to image quality, realism, depth, and immersion has taken nearly two decades to de
ecosystems.
Abstract: Modern production and distribution workflows have allowed for high dynamic range (HDR) imagery to become wide
luminance range allowed by HDR ecosystems could introduce unintended, high-magnitude changes. These luminance changes
three components of HDR luminance changes: the magnitude of the change, the direction of the change (darker or brighter), a
adaptation time results in more discomfort at the unintended transition. We find that transitioning from brighter to darker sti
Abstract: Computer simulations are presented of a neural model of lightness perception that is demonstrated to account, in a
hypothesizes that separate populations of ON and OFF neurons in the early visual pathway, having different characteristic neu
set of qualitative and quantitative findings on lightness, including lightness matches made with the Staircase Gelb (SC) and scr
Cornsweet illusion). I discuss how the model might be extended to the chromatic domain using similar biologically-inspired pri
well-known models of color and lightness, including the retinex color theory of Land and McCann and the ODOG lightness mod
Abstract: In this paper, we present two lines of research aimed at understanding how different visual, kinaesthetic, and tactil
Reality (MR), revealing superior performance in MR for all four tasks measured. In the second, we have been developing a low
photographed under UV light.
Abstract: Avoiding person-to-person collisions is critical for visual field loss patients. Any intervention claiming to improve the
pedestrians must be clearly defined. We measured nine normally sighted subjects’ collision envelopes (CE; an egocentric boun
testing platform, non-colliding pedestrians should not evade the CE.
Abstract: Recently, the traditional rear and side view mirrors have been started to be exchanged with a digital
version. The aim of this study was to investigate the difference in driving performance between
traditional rear?view mirrors and digital rear view mirrors which is called Camera Monitor System
(CMS) in the vehicle industry. Here, two different types were investigated: CMS without or with
Augmented Reality (AR) Information. The user test was conducted in a virtual environment, with four
driving scenarios defined for testing. The user test results revealed that the participants’ driving
performance using CMS (only cameras and 2D displays without augmented information) did not
improve over traditional mirrors. Most participants indicated that they would only upgrade from a
traditional mirror to a CMS car with AR, rather than just a CMS car, as CMS did not provide enough
benefits over traditional mirrors
Abstract: We study the relationship between acceptability/annoyance (AccAnn) rating and quality ratings of UGC videos. Acce
This relates to the willingness of users to use those services. While audiovisual quality estimation models have a long research
Abstract: To collect data on human perception of distortions, different subjective methodologies exist, from direct rating with
the assessed stimuli on a perceptual scale (e.g., from low to high quality). Maximum Likelihood Difference Scaling (MLDS) solv
However, one limitation is the perceptual scales retrieved from stimuli of multiple contents are usually not comparable. We p
scores in terms of precision by analyzing discriminability, efficiency by comparing with experimental effort costs, and robustne
Then, we show how discriminability in the data impacts the resolving power of popular objective quality metrics. We show tha
Abstract: Human eyes cannot detect subtle distortions in videos. The threshold at which the Human Visual System (HVS) begin
Ratio (SUR), indicating the percentage of observers unable to discern any difference between a distorted video and its referen
There are some previous studies focused on subjective assessments to determine JND and subsequently SUR. However, existin
To address these challenges, we have compiled a novel VW-JND dataset based on H.265, employing CRF as a more versatile pr
using various Video Quality Metrics (VQM) as proxies for distinct datasets. Moreover, we also conduct an analysis of the bitrat
Abstract: User expectation is one of the main factors that drives the user satisfaction for video streaming service providers and
into account provide online social media platforms with increased efficiency and users with higher satisfaction. In this work, w
video context that are typically found in online social media platforms. Furthermore, we ran extensive analyses to estimate th
and annoyance thresholds.
Abstract: For the last 30 years, multimedia quality assessment has continuously gained research efforts along with the develop
service. As any experience happens in a context (e.g. viewing conditions, expectations, cost), the impact of multimedia quality
impact of multimedia quality on QoE in sustainable context. This talk introduces a few concepts of the young QoE science field
will describe a few recent studies to measure the impact of context on both perceived quality and QoE.
Abstract: We present two novel algorithms that reduce the VR power consumption with little degradation on visual perception
the two power saving techniques. We perform human subject studies, which show that our techniques are able to save over 5
Abstract: The quality of web conferencing services often degrades by network quality. Parametric video quality-estimation tec
techniques, we have to consider display layouts such as single and grid views, which are layouts unique to web conferencing v
comprising the display and wider quality distributions.
Abstract: The transparent display has attracted attention from industrial and research fields due to its broad applications in m
conditions and cluttered backgrounds. This paper analyzes visibility degradation under various backgrounds and ambient cond
The experimental results show that the proposed algorithm can enhance visibility under various backgrounds and ambient con
Abstract: Conventional image quality metrics (IQMs), such as PSNR and SSIM, are designed to be used with perceptually unifo
consist of standard-dynamic-range (SDR) images collected in standard – possibly uncontrolled – viewing conditions. Popular pr
limited available data. In this work, we explore the use of domain adaptation to reduce the degradation in performance of dee
pretrained on SDR (source domain) images to be used for HDR (target domain) IQA. We validate our method on several popul
generalize to unseen data.
Abstract: Assessing the quality of images often requires accounting for the viewing conditions - viewing distance, display resol
large distance. Nonetheless, traditional metrics are limited when applied across a diverse range of users with a diverse range o
account for the change in the viewing distance. In a recent study comparing these two types of metrics, the authors did not ob
rescaling-based metrics. We compare the behaviour of each approach and investigate the correlation between their predictio
downscaled), but not when the viewing distance is decreased (the images are upscaled) or when accounting for low-frequency
Abstract: Advances in artificial intelligence are revolutionizing computer graphics, with generative and deep-learning methods
development of AI-enhanced content? At NVIDIA, principles of vision science guide development and evaluation of many AI-b
cover some open challenges in the area, and potential research directions to address them.
Abstract: This presentation offers a two-part insight into the realm of image quality assessment. The first part provides an ove
validity. The second part delves into the 'Psychological Image Quality' experiment, highlighting the influence of emotional con
such as psychology, data analytics, and machine learning.
Abstract: This paper proposes a novel aggregation method using the Kumaraswamy distribution to analyze partial metric value
efficacy, revealing how different video segments affect perceived quality across varied age groups. This approach, grounded in
Abstract: We created a database of ten images representative of typical digital photographic imagery, and evaluated both the
images. We then used the methodology of Liu et al. to assess a compositional figure of merit, EA, for each of the test images. W
experienced observers from the arts community, and a rank order for aesthetic merit was determined from their evaluations.
adherence of image composition to the “rule of thirds”, suggesting that it dominates sophisticated observers’ intuitive appreh
Abstract: Film grain is an artifact of analog photography caused by the silver halide process that, for a variety of reasons, movi
and video. We selected a diverse set of images and video sequences covering a variety of subjects and signal properties. We
intermediate level of noise; at low-level noise, the preference of film grain over white Gaussian noise is not statistically signific
preference for the lower amount of noise in all cases. For video sequences, we compared two levels of film grain noise to nois
Abstract: Individuals with aphantasia report either absent or dramatically reduced mental imagery compared to control partic
assess the contents of visual imagery for aphantasic individuals relative to controls. Prior analyses of the Aphantasic Drawing D
than an overall reduction in imagery might suggest. Here, we examined the mid-level image statistics of line drawings in this d
find clear differences across several different sets of mid-level properties as a function of aphantasia, which offers further char
Abstract: The ecological validity of user studies has increasingly gained attention in the field of Quality of Experience (QoE). Ho
of existing guidelines and main aspects to consider when aiming to design and conduct more ecologically valid studies in the Q
In this presentation, we present a set of guidelines that provide guidance towards conducting ecologically valid QoE user studi
Study Design, Task and data collection) overarching three main dimensions (Setting, Users and Research Methodology).
We further reflect on the broader implications of ecologically valid user studies for QoE research and provide a basic visualisati
Abstract: The most computationally modelled topics of vision science have built on signal detection theory. Further, this aspe
added to a noise, and the visibility or appearance measured. Over time the parameters of the signal were elaborated from sim
filtered noises, then noises of different histogram shapes (uniform, Gaussian, etc.), again both in the achromatic and chromati
a reverse manner, where distortions (noise) were treated as signals and the imagery as a noise (masker). Image distortions of
distortions such as ringing, blocking, frame rate and, perhaps surprisingly, distortions such as blur, color and tone scale. Now
entropy and has advantages in terms of comfort noise. But the SKE (signal known exactly) paradigm of signal detection theory
often contains shape distortions. One key category is perspective where subtle imperceptible distortions result in large contra
(visually lossless) and no-reference (visually plausible) gets challenged. This talk will go into detail on the new types of experim
Abstract: Pixel designs can be made in a variety of ways to produce full color. Designs using stripe, S-stripes, pentiles, and hex
pixel design, we created various stimuli. They are stripe line-patterns that are oriented in five distinct directions (angle: 0, 90, 4
shape is lower than that of other designs in all directions. This implies that hexagonal shapes with comparatively lesser resolu
stimuli (such as text, words or images utilized in daily life) are required in the future.
Abstract: The state-of-the-art displays have became bigger and brighter. Since it was revealed that brightness and size affect fl
with 30Hz, we observed the brightness effect and size effect on display flicker perception. Participants perceived flicker strong
effects of brightness and size.
Abstract: Seabed 2030 is a collaborative project between the Nippon Foundation of Japan and the General Bathymetric Chart
My work centers around this newfound data and has already revealed a glimpse of the secrets hidden from view on the seaflo
"As above, so below."
Abstract: In this study, we designed an experiment using remote-controlled lab based moving platform to evaluate the impact
point and parking it between two boxes. Participants provided feedback through questionnaires, and their experiences were a
well as a simulator sickness questionnaire. There was an impact of resolution on the perception of the user as tasks were exp
resolution. The latency was also influencing the user in their performance as well as their experience of the platform. The field
Abstract: Vision Transformers have emerged as powerful alternatives to convolutional networks, however, they require much
target. We explore the question: is it better to adapt the source-trained teacher and then distill to the student model, or to dis
performance to the larger network.
Abstract: Wood defects detection and classification are critical tasks in the wood industry, as they directly impact product qua
information in real-time using deep learning and tracking algorithms, even in scenarios with multiple cameras capturing differe
(SORT) algorithm is used to track planks containing detected defects as they move through the frame. Additionally, the metho
defects across different camera views. The proposed method has been tested on a dataset of wood planks, demonstrating hig
Abstract: In this paper, we address the task of detecting bees inside a beehive task using computer vision with the goal of mon
we propose to install camera in the beehive and analysis with deep learning techniques. However, the lighting conditions acro
detection on Deep Neural Networks. In this paper, we propose to apply domain randomization based on motion to detect bee
detection, especially for small bees which are more likely to be affected by the lighting conditions and misclassified by the bee
Abstract: In this paper, we introduce a unified handwriting and scene-text recognition model tailored to discern both printed a
significant advantages: 1) A substantial improvement in the recognition accuracy for both scene-text and handwritten text, an
Abstract: In this paper, we present a deep-learning approach that unifies handwriting and scene-text detection in images. Spe
Furthermore, we build a new benchmark dataset that comprehensively captures various handwritten and scene text scenarios
Abstract: Intelligence assistance applications hold enormous potential to extend the range of tasks people can perform, increa
techniques based on massive foundation model networks cannot run on devices at the edge. A remote intelligence server can
models. We argue that a hybrid ensemble architecture that combines intelligent switching of efficient focused networks and a
identify mistakes in complex procedures and provide reasoning about mitigations. Since system inferences are still imperfect,
dramatically improves the usability of the system. Importantly, our interface provides feedback without compromising situatio
Abstract: The utilization of dual-energy X-ray detection technology in security inspection plays a crucial role in ensuring public
the quality of the displayed image and affects the performance of the automatic threat detection pipeline. While deep learnin
security inspection scenarios. This limitation hampers the widespread application of these methods in the field of security insp
clean reference images for training. Our denoising approach is built upon the Blindspot neural network, which effectively addr
that our method achieves favorable BRISQUE scores across different baggage scenes.
Abstract: In this paper, we study the problem of image restoration from highly incomplete data, in formulation of compressive
denoisers. Nonlocal group sparsity and self-similarity priors together with deep learning are used in this framework. A thoroug
introduced methods are outperform existing methods.
Abstract: Solving the problem of facial video inpainting holds significant potential benefits for various domains. This research p
method distinguishes itself from existing methods by prioritizing the recovery of identity, intricate facial details, and emotiona
satisfactory outputs. This advancement is essential for enhancing user experiences across multiple applications, benefiting the
Abstract: Remote photoplethysmography (rPPG) is a camera based technique used to estimate a subject's heart rate from vide
human eye, these subtle changes are easily detectable from high-quality video. Working with high-quality video presents man
through the use of compression algorithms, but modern compression algorithms are unconcerned with maintaining the small
information and may predict heart rates less accurately. We examine the effects of compression on deep learning rPPG with m
video still returns a valid heart rate. These compression techniques are applied against multiple public rPPG datasets (DDPM a
H.265, and at a constant-quality mode of CRF above 37 for VP9. We find that training on compressed data yielded less accurat
Abstract: Although CNN-based classifiers have been successfully applied to object recognition, their performance is not consis
cases, the poor performance of ill-performing classes is due to biased training samples, which fail to represent the general cov
results show some promising results.
Abstract: This talk highlights the cutting-edge developments and practical applications of lensless phase microscopy, a revoluti
improved algorithms to machine learning integration, showcasing its expanding utility. Diverse applications, such as cellular im
Abstract: Spectral rendering encompasses methods aimed at generating synthetic imagery with realistic color expression by si
spectrum of light as input, which allows for more accurate color reproduction and more realistic visual effects. However, spect
problems in spectral rendering, covering the main stages of the rendering pipeline such as RGB to spectrum upsampling, spect
elaborate overview of the current state-of-the-art and to identify future research directions and opportunities in this exciting
Abstract: There are many electronic documents salient to read for each given topic; however, finding a suitable reading order
suitable and optimal reading order for curriculum generation quantitatively. It is necessary to read the relevant documents in
ground truth. Our method suggests the best reading order automatically by checking the relevant topics, document distances,
using word frequency, and topic sets. We measure the similarity, relevance, distance, and overlap of different documents usin
documents. We evaluated the performance of your system against the ground truth reading order using different kinds of tex
Abstract: In this paper, we present a new framework for graph-based semi-supervised learning. We present a powerful metho
framework focuses on exploiting the data structure and soft labels of the available unlabeled samples. The proposed method
performance compared to semi-supervised methods that use integration and label inference simultaneously.
Abstract: In the dynamic realm of image processing, coordinate-based neural networks have made significant strides, especial
memory challenges. Addressing these, this study introduces an innovative approach using Tensor-Product B-Spline (TPB), offe
to sidestep computational burdens of neural fields. This was achieved by replacing iterative processes with deterministic TPB s
optimize image reconstruction. Evaluation on the Set14 and Kodak datasets showed the TPB-based approach to be comparab
perspective in image processing, setting a promising trajectory for future advancements in the domain.
Abstract: Deep learning has enabled rapid advancements in the field of image processing. Learning based approaches have ac
parallel development of hardware architectures capable of optimizing the inferencing of deep learning algorithms in real time
deep learning algorithms. Thus, associated systems have to be evolved to make use of the optimized hardware balancing com
multidimensional data to optimize inference time of deep learning algorithms for vision applications.
Abstract: Light field displays have a finite angular and spatial resolution, which limits the display's depth-of-field, that is the de
appear distorted due to aliasing. This aliasing can be mitigated by properly blurring those parts, with blurring preferably done
paper we propose a method for simultaneously rendering multiple adjacent views in a light field, with each of them having the
single view without compromising on the overall rendered quality.
Abstract: There is increasing evidence that standard sharpness (MTF) measurements correlate poorly with Machine Vision and
We describe new techniques for measuring noise in the presence of slanted edge signals that enable the calculation of the key
We expect information capacity to be a strong predictor of MV/AI system performance, and because it is relatively unaffected
result in the fastest calculations and least power consumption.
The most important of the additional metrics, SNRi and Edge SNRi, measure the quality of object and edge detection, which ca
Finally, we discuss the mathematical framework that ties the new metrics together, resulting in a powerful and versatile toolk
Abstract: This paper seeks to investigate the relationship between image quality and Computer vision performance. Two imag
characterizes sharpness in images and 2. Contrast Transfer Accuracy (CTA) which characterizes contrast in an image. Compute
shown to exist between MTF and OD performance, while CTA is shown to be a flawed predictor of OD performance. These re
performance.
Abstract: Single-photon pulse LiDAR has a wide range of applications in navigation and object detection due to its precise rang
analyzes the photon arrival statistics, and derives the mean squared error (MSE) as a function of the number of sensors in a un
Abstract: Event-Based Vision Sensors (EVS) utilize smart pixels capable of detecting on a pixel level whether relative changes o
conventional CMOS Image Sensors (CIS). Emerging hybrid EVS+CIS sensors target fusion of the high image quality, high spatial
Blurred Edge Width Metric (BEW) to benchmark EVS assisted slow-motion capture against CIS-only solutions. The EVS assisted
VFI solution. Furthermore, it is shown that the hybrid EVS-CIS based VFI slow-motion capture achieves similar performance as
Abstract: Consumer cameras have become indispensable tools for communication, content creation, and remote work. Qualit
systems perform in face-present scenes containing diverse skin tones, and how performance can be objectively measured usin
exposure (AE), and color reproduction.
We assess the performance of several consumer cameras (web, phone, DSLR) with different lighting conditions and skin tones
exposure, and reduced overall image fidelity are unfortunately common for scenes containing darker skin tones, revealing a m
Moreover, we present a comprehensive evaluation of current standard procedures, highlighting strengths and limitations in m
range of users and environments.
Abstract: Portraits are one of the most common use cases in photography, especially in smartphone photography. However, e
on realistic mannequins, a laboratory setup that can cover all commercial cameras from video conference to high end DSLRs. O
the input image 3) the merge of the two quality scales to produce the final wide range scale.
On top of providing a fine-grained wide range detail preservation quality output, numerical experiments show that the propos
Abstract: Portraits is one of the most common use cases in smartphone photography, however rendering a high quality portra
not the preferred skin tone. Moreover, user preference studies have been developed mostly in the laboratory, a simplified sce
Abstract: The importance of low-light defect detection for image sensors is increasing. This paper presents a deep learning-ba
with different depth. Then with strategically constructed large-scale datasets, we successfully trained general models that ach
method. Additionally, the regression model is in good agreement with the qualitative evaluation result from 30 expertise and
Abstract: Ever since the advancements of high-performance processors in mobile phones, complex image signal processors (IS
the image quality assessment (IQA), which verifies the ISPs, is becoming increasingly important to sensor manufacturers. In th
pseudo reference image by detecting distorted pixels using an outlier detection method with the Gaussian mixture models an
our proposed metric, we conducted a qualitative evaluation survey, and its Pearson correlation coefficient (PCC) was calculate
Abstract: TBD
Abstract: In this paper, I will present the results of a virtual reality subjective experiment that will be performed this semeste
experiments are being performed as part of the VQEG- IMG test plan for the definition of a new recommendation for subjectiv
application and detail the VQEG-IMG test plan tasks for XR subjective QoE asssessment. I will also describe the experiment to
Abstract: With the fast-evolving video and display technologies, there is a growing need to better understand user preference
influence viewer experience. We conducted two psychophysical experiments to investigate the latent factors affecting perceiv
researchers and display designers aiming to improve display and environment settings to give end-users an optimal viewing ex
Abstract: While polarizer-free Pol-less OLED displays increase the efficiency of the panels while with lowering the thickness, th
features of the diffraction appearance were classified into the size and complexity of the pattern, and each factor can be auto
stage.
Abstract: An accurate and repeatable measurement of lens flare has been created at DXOMARK. The same method and metric
level. This limitation proved to be problematic for automotive cameras with integrated ISP with no access to RAW images. It a
parameters.
This new measurement gives the exact same metrics “flare attenuation” in RGB and in RAW. It uses a new measurement setup
Abstract: The recently published ISO 12233:2023 has introduced a new edge-based spatial frequency response (e-SFR) feature
between sagittal and tangential SFR with vertical and horizontal SFR across the field of view. However, the extended utility gra
resolution across these orientations. This paper provides recommendations for the efficient and accurate detection and analys
Abstract: In the field of endoscopy, the growing focus on early cancer detection drives technology advancements. Hyperspect
upcoming submissions. However, the absence of systematic HySE performance assessments and standardized evaluation met
Our project develops two adaptable HySE prototypes employing spectral scanning and snapshot techniques. We target essenti
bands. This endeavor uncovers vital HySE performance factors and their evaluation methods, streamlining assessment of subm
Abstract: The capability for a camera to produce a color-accurate high dynamic range image is based upon its capture of lumin
appropriately capture an object’s appearance. Color accuracy is critical to documenting cultural heritage appropriately. This re
extended reflectance dynamic range, color reproduction quality of high dynamic range scenes, and generation of the high dyn
12 Pro, and two illumination levels approximately 800lux to 57lux with a wide range of color targets with various surface textu
high dynamic range images.
Abstract: In photography, the dynamic range (DR) is a distinguishable brightness range and is determined by the analog-to-dig
However, since camera manufacturers set these HDR algorithms to operate differently by considering the situation, it is neces
the SNR of the picture. However, it is difficult to measure the two information in general-scene photos without charts. To over
reaches 12dB. Using the pre-calculated radiation and SNR information, we measured DR of photos without using a chart, and
Abstract: This paper provides elements to answer the question: how to judge general stylistic color rendering choices made by
scene, supporting both HDR and SDR content. To that end, we discuss modeling of camera behavior and visualization methods
Abstract: This paper presents a method for synthesizing 2D and 3D sensor data for various machine vision tasks. Based on a 3D
arrangements can be generated automatically. In addition, objects can be virtually deformed in order to create realistic image
applications can be created, even in the absence of defect parts.
Abstract: PCB defect segmentation aims to localize the defects in printed circuit boards (PCBs). While this problem has great in
with various defective categories of imbalanced distribution, reflecting real world conditions. The problem of unsupervised PC
Specifically, a pretrained large foundational model (SAM) is adapted to PCB-UAS by introduction of a few learnable adapter lay
images that mimic the real ones. Experiments highlight that the proposed method can outperform baselines by 7 points with 1
Abstract: Accurate measurement of daily feed consumption for dairy cattle is an important metric for determining animal hea
non-invasive analytics system that leverages depth information derived from stereo cameras to consistently measure feed off
feed volume from the 3D projection. Additionally, an occlusion detector complements our system to produce occluded volum
optimizing feed management in dairy farms, thereby improving animal health and sustainability in the dairy industry.
Abstract: An Erdstall is an underground passage or tunnel, often dug in sandy or loose soil. These structures are found in vario
Erdstalls is not clearly understood and there are various theories about their use. Some researchers believe that they were use
places for certain craft activities. Mounds are often difficult to date as there is limited archaeological evidence to determine th
narrow environments and at the end to map the environment in 3D. Creating a 3D map of Erdstall facilities can serve various p
Abstract: High dynamic range (HDR) image sensors have become increasingly important for automobile applications, particula
image sensors in the industry. This study used a monochrome CMOS image sensor to model single and split photodiode pixel i
function (MTF) from the split photodiode. MTF of the large photodiode in the split photodiode pixel at half Nyquist was lower
traditional single photodiode pixel does not experience a change in MTF with light level and edge artifacts were not observed.
Abstract: We present a framework to model the energy consumption of emerging computational image sensors. The model is
specifically for gaze tracking in Virtual Reality. We highlight several design decisions in building the sensor to reduce the overa
Abstract: This publication presents a novel single slope single ramp column analog-to-digital converter (ADC) for CMOS image
for the reset level. To compensate for offset mismatch, a common reference level is used to start the conversion, while the re
correlated double sampling (CDS) output is achieved by subtracting the reset value from the signal value with a digital subtrac
noise. A 12-bit column ADC was designed and implemented in a 1920 x 1080 matrix resolution CIS demonstrator, using a 0.18
Abstract: Accurate coupling capacitances are a key component in the design of modern image sensor cells. Automation of 3D
boundary conditions. High structural complexity demands efficient numerics for reasonable runtimes. We compare capacitanc
7 interconnect layers in a few minutes on a standard Linux machine.
Abstract: Ideally, a compelling virtual or augmented reality display should “pass the visual Turing test”, in other words, be so r
dependent effects in a compact form-factor. However, despite their potential, holographic displays have yet to become mains
at addressing these issues.
Abstract: Recently, many deep learning applications have been used on the mobile platform. To deploy them in the mobile pla
networks.
In previous study, we studied the effect of the quantization of activations and weight for deep learning network on image qua
In this paper, to achieve more quantization bit reduction than previous work, we made adaptive bit-depth control of input pat
With proposed method we could achieve 5% reduction in hardware area and power consumption for our custom deep netwo
Abstract: The design of camera spectral sensitivity (CSS) is important for accurate color reproduction. However, it is known tha
associated color correction matrix (CCM). Regarding this tradeoff, Tan et al. introduced a metric that evaluates the color repro
against noise but has lower color reproduction accuracy [2]. Inspired by those works, we propose a joint CSS and CCM optimiz
sensitivity of each wavelength should be within the range of zero to one. Then, we jointly optimize the CSS and the CCM by a p
outperforms an existing CSS.
Abstract: The integration of trench vertical transfer gates to an indirect time of flight pixel has been achieved through TCAD &
(QE) and power consumption. A small fast photo-gate pixel design surpassing state of the art performances has been validated
Abstract: This paper presents an extension of the dynamic range of a short-pulse indirect Time-of-Flight (ToF) range sensor wi
output range to be optimized by controlling the signal accumulation time at each subframe. The dynamic range is extended w
region, a technique of restoring entirely linear signal from linear-logarithmic response using the proposed theoretical formula
scale range for the case that the dynamic range is extended by 30 times using a logarithmic response. The proposed technique
greater than the maximum light.
Abstract: This paper proposes a framework of pixel-wise parameter estimation for Event-based Vision Sensor (EVS) characteriz
parameter estimation is then formulated as an optimization problem to minimize the measurement-prediction error for both
S-curves.
Abstract: Imaging dynamic scenes accurately and efficiently is vital in domains such as computer vision, medical imaging, and s
challenging because of the decreased signal-to-noise ratio (SNR) caused by the short exposure time in each frame. Quanta Ima
dynamic scene imaging pipeline that combines low-bit QIS with a new video restoration algorithm. We demonstrate 3-bit QIS
We propose an end-to-end REcurrent QUanta VIdeo REstoration (QUIVER) network to reconstruct high-frame-rate videos from
exploration of low-bit QIS's capabilities. We train the proposed methods using the I2-2000FPS dataset, and experimental resul
Abstract: The Magic Lantern project describes itself as "Magic Lantern is a free software add-on that runs from the SD/CF card
on many Canon EOS interchangeable lens cameras, and also useful well-documented interchange formats for data extracted v
The current work describes how these facilities can be applied by researchers to develop new imaging techniques on a profess
Specifically, this work covers an to use the Magic Lantern development tools to manipulate the Embedded Direct Memory Ac
sensor of a prosumer camera as input for alternative processing pipelines.
Abstract: We use data-driven approaches to solve ill-posed inverse problems that frequently arise in engineering design. For e
reconstructed density fields. As such, in this setting, rather than use the full radiographic image, we focus on identifying featur
parameters. Furthermore, rather than use a purely computational-imaging based approach to inversion of density from radiog
based forward model integrations to generate learning data (cf. transfer learning). In the synthetic data setting we examine, w
networks and variational autoencoders.
Abstract: In the realm of digital camera simulation, the lens plays a pivotal role, acting as an intermediary between the scene a
presented as black boxes, obfuscating details like material properties, wavelengths settings and refractive indices. A significan
data for multispectral optics simulation. This paper introduces a systematic methodology comprising stages like data preparati
multispectral processing. By achieving this, we aim to enhance the fidelity and accuracy of digital camera simulations, offering
Abstract: A trained attention-based transformer network can robustly recover density fields from a sequence of features deriv
transformer encoder, which acts on a sequence of features extracted from noisy radiographs. This encoder includes numerou
which is greatly obscured by the radiographic noise, can be reconstructed accurately, despite the low dimensionality of the fe
Abstract: Characterization of material properties of objects undergoing strong deformations is an important task in material sc
probabilistic neural network, trained with a database of multi-physics simulations of the physics problem in question, to map p
then map the predicted population of parameters to features using a deterministic network and find that the features are rec
Abstract: A challenging problem in industrial radiography is accurate density reconstructions from X-ray projections corrupted
Abstract: In this paper, we introduce silhouette tomography, a novel formulation of X-ray computed tomography that relies on
that any solution exists. We then propose a supervised reconstruction approach that uses a deep neural network to solve the
Abstract: Numerous clinicians routinely generate reports from ultrasound (US) scans. Due to the real-time nature of US imagin
for US image report generation using a Large Language Model (LLM) chain tool. This tool autonomously considers user input a
classification, to enable the generation of reports from breast ultrasound images.
Abstract: The reconstruction of 3D Cone-beam Computed Tomography (CBCT) images from a limited set of projections is an im
regularization is sub-optimal when the number of available projections is limited. In the past decade, deep learning (DL) has ga
slice-by-slice manner or a 3D CNN that performs the whole volume reconstruction via 3D convolution. However, the suboptim
Sparse-view 3D image reconstruction. Swap-Net produces 3D volume reconstruction in an end-to-end fashion without using fu
Abstract: In a broad range of radiographic applications, 3D shapes must be reconstructed from 2D projections. CNN-based app
target 3D shape, and, while their complexity can be overwhelming, it is important to keep in mind that, by Whitney's theorem
features with an autoencoder network.
Abstract: We propose a modification to the classic difference of Gaussians (DoG) and scale-invariant feature transform (SIFT) a
summed area tables and box filters, modality invariant key points are efficiently identified as local extrema in the change in th
materials science.
Abstract: This work seeks to collect SEM, BSE, and Scanning White Light Interferometry (SWLI) images of Ti-6AL-4V fatigue frac
rate. SEM images are the standard for studying the topography of fracture surfaces, but BSE images and SWLI data offer the a
images and variety of other media. CNNs have been used to make successful Classification and predictions of fracture surfaces
Abstract: Three-dimensional electron back-scattered diffraction (EBSD) microscopy is a critical tool in many applications in ma
the opportunity to use transformers, state-of-the-art deep learning architectures that have made breakthroughs in a plethora
EBSD volume, using an efficient transformer model and a projection algorithm to process the transformer's outputs. Overcom
superior recovery accuracy on real 3D EBSD data, compared to existing methods.
Abstract: 3D imaging of biomolecular dynamics is essential for understanding how it functions in living systems. Current 3D im
method that recovers a molecule’s 3D dynamics from liquid-phase EM movies. We demonstrate the performance of TEMPOR
Abstract: We present a fully differentiable hydrodynamics framework to facilitate the recovery of hydrodynamic code parame
differentiation
Abstract: DCE-MRI provides pharmacokinetic parameter-based insights into vascular permeability and tissue perfusion. Conve
helps, most approaches depend on paired DCE-MRI and labeled pharmacokinetic maps, hampering practicality due to labeling
generator and discriminator, outperforming traditional techniques by yielding more reliable results without separate AIF meas
Abstract: Laser powder bed fusion is a technique in additive manufacturing which uses a laser to selectively
fuse layers and produce components from a bed of powdered material. Despite inherent benefits to
engineering and design, adoption of the technique has been slowed by the ongoing prevalence of
fabrication failures most often caused by spatter (ejection of molten material) with certain metals and
alloys. This work focuses on locating spatter in the powder bed environment through the analysis of
image data and performing semantic segmentation using vision-based machine learning. With
collaborative efforts from an expert in the field, a series of feature masks were created to train a
supervised learning model and variations of the same model architecture. Performance was
quantified using the Intersection over Union (IoU / Jaccard Index) metric and recording the time
required to perform inference on a single sample.
Abstract: For nonlinear inverse problems that are prevalent in imaging science, symmetries in the forward model are common
difficulties arise, and more importantly, how to overcome them by preprocessing the training set before any learning, i.e., sym
driven learning performance. We also formulate the principle of symmetry breaking that can lead to efficient learning.
Abstract: Starch plays a pivotal role in human society, serving as a vital component of our food sources and finding widespread
granules. Quantitative analysis through the segmentation of starch granules from the background aids researchers in explorin
with remarkable accuracy. The method yields impressive results, with average values for several evaluation metrics—namely,
Abstract: Abstract: ACES is a standardized color management system widely used in the film and visual effects industry to ensu
developers who want to achieve high-quality, consistent color representation across different platforms and displays.
In current high-end media pipelines, there is a strong tendency to combine film technologies like Visual Effects and real-time p
installed using the Unreal game engine as the content player. This technology enables filmmakers to change the displayed con
By leveraging ACES in Unreal Engine 5, game developers can achieve heightened visual fidelity, especially with respect to wide
devices and display technologies. Moreover, Unreal Engine 5's support for ACES facilitates seamless collaboration with other c
However, implementing ACES in a real-time engine presents unique challenges regarding performance optimization and ensur
content.
This paper uses ACES to investigate color input and output consistency to and from Epic Games Unreal 5 regarding Wide Color
The topics are:
· ACEScg as a working colorspace in Unreal 5
· Texture workflow from Megascan Library
· Export to the Foundry Nuke
· Export to Blackmagic DaVinci Resolve
· Lighting Consistency
Abstract: The JAB Code, a 2D barcode standardized in ISO/IEC 23634:2022, offers improved reliability and data capacity over t
conversion, prompting the need to select colors that ensure distinct segregation in the transformed color space for enhanced
We propose an approach for calibrating the colors of the JAB Code, involving the creation of a test pattern, quantization of the
generation tools or web apps, simplifying the process for users and ultimately improving color accuracy and fidelity within the
We conduct an experiment with different printers, utilizing a smartphone for image capture. The evaluation includes printing
analyze color distances in the RGB space, revealing improvements in color distribution and lower error rates with printer color
next revision of the ISO standard.
Abstract: Abstract: Sustainable tourism has emerged as a critical imperative in the face of environmental challenges and incre
edge technology and user engagement, these digital posters aim to foster a deeper connection between tourists and the natu
The proposed methodology combines augmented reality (AR) and geospatial data to design interactive trails that present imm
delve into the significance of each site while learning about sustainable practices.
By bridging the gap between traditional signage and modern technology, these posters serve as captivating educational tools
insights, facilitating further improvements to sustainable tourism initiatives.
Ultimately, this innovative approach can revolutionize the way tourists engage with their surroundings and contribute to the p
Furthermore, the development and implementation of these interactive trails address several challenges faced by sustainable
devices. This empowers tourists with valuable knowledge about the local environment, heritage, and responsible travel practi
Secondly, the incorporation of geospatial data ensures that tourists can easily navigate through the trail, minimizing the risk o
and reducing environmental footprint.
Moreover, the interactive nature of these rich media posters fosters a sense of engagement and interactivity, making the tour
and recommendations with others, further promoting sustainable tourism practices.
As tourism plays a significant role in global economies, the urgency to safeguard the environment and preserve cultural herita
While this article highlights the potential of this approach, further research and pilot projects will be essential to refine and va
Abstract: Abstract The Driver Monitoring System (DMS) presented in this work aims to enhance road safety by continuously m
providing real-time alerts to mitigate potential hazards. The primary components of the DMS include gaze detection, emotion
deep learning models to analyze the driver's facial expressions and extract dominant emotional states. In case of detected em
integration of these features creates a comprehensive driver monitoring solution that assists in preventing accidents caused b
Abstract: ABSTRACT
The research aims to investigate the influence of tweets from two prominent figures in the technology industry, Elon Musk, an
significant correlations between specific tweets and fluctuations in stock prices, thereby shedding light on the role of social me
Historical information on the daily stock prices of Tesla and Microsoft, as well as tweets by Elon Musk and Bill Gates, were take
tweets.
The findings of the study reveal the potential influence of tweets by Elon Musk [7] and Bill Gates on the stock prices of Tesla a
regulators in understanding the implications of social media activities on stock prices and investor behavior.
Abstract: Abstract: This elaboration presents a prototype that aims to make education accessible to all. The chosen learning to
review and the development of a prototype, which was tested using the BITV tests. The key findings are that accessibility shou
elements of software are not necessarily needed for its core functions. With the insight of using an agile participation model, a
Abstract: In 2021, there were 1,157 general aviation (GA) accidents, 210 of them fatal, making GA the deadliest civil aviation c
difficult to map into a real 3D environment. To address these issues, Augmented Reality (AR) was utilized to provide 3D immer
on a mobile device especially challenging for complex visualization of weather phenomena. This paper presents research on h
technological evaluation of the application's rendering and tracking performance across different devices is presented.
Abstract: Abstract: Critical infrastructure is the backbone of modern societies, and protecting this infrastructure is essential to
this sector is constantly evolving. With the increasing sophistication of cyber-attacks and other threats, it has become essentia
Open Source Intelligence (OSINT) technologies have emerged and offer valuable tools for identifying and mitigating these thre
This article presents an in-depth overview of OSINT technologies and their applications in the protection of critical infrastructu
successful applications of OSINT technologies in the electricity sector are presented to illustrate their effectiveness.
This article also examines organizations' challenges in implementing OSINT technologies, including technological, legal, and fin
Finally, the article concludes by offering recommendations for successfully implementing OSINT technologies to protect critica
infrastructure.
Abstract: Abstract: The fast growth of cloud computing technology has led to immense development in the public and private
public to advance with this technology. However, they cannot assuredly place their information in the cloud due to many secu
Moreover, in the scope of cloud computing, the importance of security testing must be considered. Security testing involves e
methodologies for cloud security testing. Furthermore, the paper introduces the tools offered by three significant CSPs for clo
Azure offerings for securing their clouds and some known tools for security testing in the cloud. Lastly, we introduced and exp
Abstract: Abstract: With the widespread use of social media, current times have noticed a huge increase in the number of cyb
compromised accounts for their personal gain before they are recovered. Once the account is recovered, it is a tedious task to
activity done by an attacker on a compromised social media account. In this tool we plan to use natural language processing a
and recovery was reported respectively.
Our proposed tool will also use advanced cybersecurity methods to ensure data privacy of concerned users. Furthermore, a co
knowledge, our tool doesn't just help recover from attacks quickly but also discourages cybercriminals from using social media
Abstract: Abstract: With the popularization of the internet among the masses digital currency has also seen its fair share of fam
more avenues for the actors to perform malicious activities. With almost negligible restrictions implied on them, it has becom
only provides pseudo-anonymity to the users but also keeps all the transactions transparent. In this paper, we have discussed
tools, novel techniques, and methodologies that can help manage incidents of illegal transactions.
Abstract: Many of Deep Neural Network (DNN) models are designed for high power and cloud inference scenarios. These mod
And-Excitation, SiLU, Swish, HSwish, GeLU, Focus etc. This work focuses on methods that can be used to easily remove or repl
replace SiLU with ReLU, Focus layer with Convolution layer etc.
After Model Surgery, the model can be trained using the regular model training scripts. We have trained several such models a
embedded processors – this would not have been possible without having such a Model Surgery method.
We provide details of accuracy and inference time of the original models and lite models in the results.
Abstract: The stock market's volatile nature makes accurate forecasting challenging. While traditional models like ARCH, ARMA
and ARIMA have been used for predictions, they struggle with the market's complexities. Machine and Deep Learning,
particularly neural networks, offer new avenues for prediction. Transformer models, originally for language processing,
have shown promise in stock market forecasting. This research studies three such models: Time Series Transformer,
Informer, and Autoformer. Despite their potential in time series forecasting, these models aren't yet the ultimate solution
for market predictions, emphasizing the need for continuous research and understanding of their limitations.
Abstract: As the digital landscape continues to expand, so does the complexity and scale of cybersecurity threats. Traditional c
novel paradigm of AI-based Cybersecurity Management Consulting (AI-CMC) that promises to be a disruptive technology for th
AI-CMC integrates advanced artificial intelligence techniques with cybersecurity management consulting practices, providing o
predictive analytics, and intelligent decision-making capabilities. It harnesses the power of vast data repositories, continuously
This research explores the core components of AI-CMC, including its data-driven approach, adaptive learning models, and coll
understanding of its implementation on a global scale. Through case studies and empirical evidence, we demonstrate how AI-
Abstract: In today's cybersecurity landscape, safeguarding information systems has become a top priority for businesses. Thes
vulnerability management, focusing on open-source tools for domain and subdomain enumeration, vulnerability scanning, and
option for organizations. Examining domain and subdomain enumeration tools reveals their crucial role in mapping an organiz
Nuclei, Google Dorking, Google OSV Code Scanner, Nikto, and SQLmap) demonstrates their effectiveness in identifying critical
significance of vulnerability remediation through patching, hardening, and exposure management. These processes close secu
and limitations regarding the false positives and accuracy of open-source tools for vulnerability management, empowering org
protect valuable assets.
Abstract: ABSTRACT
This paper delves into the burgeoning role of machine learning (ML) in the domain of renewable energies. As the global shift t
hydropower, solar power, and wind power, serve as the focal points of our exploration. We further investigate the implement
review of related works sets the foundation for understanding the current landscape. The crux of our study lies in the applicati
We critically assess whether the integration of ML is indispensable or merely supplementary in the drive toward maximum ren
Abstract: Abstract:
Open-source technologies (OSINT) and Social Media Intelligence (SOCMINT) are becoming increasingly popular with investigati
public and private data available is rising rapidly. OSINT and SOCMINT technologies use sophisticated techniques and special t
This work aims to find descriptive information by using the OSINT tools available on the internet. The target will be achieved w
can be made to make them more descriptive for analysts.
Abstract: In the context of digital image/video watermarking, there exists a class of techniques that rely on insertion of signal
when constructing affine transform estimators. Addressing the problem of template synchronization with Convolutional Neura
results in subpar performance. In this work, we show for the first time, that it is possible to address image template synchroni
Abstract: Captchas are used on many websites in the Internet to prevent automated web requests. Likewise, marketplaces in
of darknet marketplaces.
In this work we focus on the darknet and provide an overview about the variety of captchas found in existing darknet marketp
With our solvers we were able to achieve accuracies between 65% and 97% which significantly improved our ability to collect
Abstract: Training of convolution neural network (CNN) needs extensive use of hardware, time, and technical intelligence to su
model development and deployment. Hence there is need of protection of Intellectual property of CNN weights and also ther
In this paper a novel method for protection and identification for intellectual property related to CNN weights are presented.
Abstract: With the advancements made in the field of artificial intelligence (AI) in recent years, it has become easy to create fa
Since such face swapping manipulations are nowadays commonly used for creating and spreading fake news and impersonatio
In the past, several methods have been proposed to detect deepfakes. At the same time, new synthesis methods have also be
addition to generalizability, the robustness of these methods is also evaluated, highlighting problems and opportunities for im
Abstract: A new algorithm for the detection of deepfakes in digital videos is presented. The I-frames were extracted in order t
eyes, nose, mouth, and face frame were analyzed separately. From the Discrete Cosine Transform (DCT), the Beta component
determine the nature of the video under analysis.
Abstract: The impressive rise of Deep Learning and, more specifically, the discovery of generative adversarial networks has rev
sensitive. Furthermore, free access to forgery technologies is dramatically increasing and very worrying. Numerous methods h
detection in videos, based on residual signal analysis is presented in this paper.
Abstract: Video DeepFakes are fake media created with Deep Learning (DL) that manipulate a person's expression or identity.
flow models to capture this temporal aspect, but they are computationally expensive. In contrast, we propose using the relate
approach is effective and has minimal computational costs, compared with per-frame RGB-only methods. This could lead to ne
Abstract: In this article, we study the properties of quantitative steganography detectors (estimators of the payload size) for co
image content, the estimator may not react to embedding. With increased payload size, it starts responding as the embedding
two critical payloads as well as the estimator bias depend on image content and payload, we study a maximum likelihood estim
model to demonstrate that the cover bias can be caused by a small number of “outlier” pixels in the cover image. This is also c
Abstract: We investigate how much information a steganographer can communicate at given level of statistical detectability o
allocation strategy (informed by Warden's detector) allowing her to communicate super-square root law payload with the exa
and thus make practical use of her source.
Abstract: Recent studies show that scaling pre-trained language models (PLMs) can lead to a significantly improved model cap
capable of generating human-like text based on context and past conversations. It is demonstrated that LLMs have impressive
secret data into an innocent cover for covert communication. Our purpose is not to combine an LLM into an already designed
steganography by itself, which is defined as prompting steganography and may be a new paradigm of steganography. We will
reduced by prompt engineering, which may shed light on further research.
Abstract: There are advantages and disadvantages to both robust and cryptographic hash methods. Integrating the qualities o
comparison between robust and cryptographic hashes. Therefore, when incorporating robust hashes into cryptographic hashe
necessary to anticipate the hash bits that are most susceptible to modification, such as those that are affected by JPEG compr
Abstract: PhotoDNA is a widely used method for generating robust image hashes. It is widely used today for the detection of C
that allows the comparison of these images to be performed significantly more efficiently. We also show that both robustness
An advantage is that the existing hashes can be converted directly, so no new calculation of hashes from reference images is n
Abstract: Handwritten document analysis is a specialized field in forensic science that meticulously examines intrinsic docume
words, and character sizes using image processing and deep learning. Our unique algorithm requires forensic expert involvem
method outperforms state-of-the-art techniques.
Abstract: Advances in AI allow for fake image creation. These techniques used to fake mammograms. This could impact patien
to identify the machine that created any mammogram. We analyze mammograms of 1574 patients obtained on 7-different m
dataset.
Abstract: In this paper, we present a method to secure sovereign documents. It is based on technical guidelines from the Inter
metadata and facial image. The data is digitally signed and stored in a JAB Code, a polychrome barcode, and printed on the so
means of his smartphone. The evaluation of the implementation was performed on a generalized concept for sovereign docum
Abstract: Real-world optical aberrations in light field (LF) displays can significantly influence overall display performance. They
In this study, we employed an optical model representative of a near-eye, time-multiplexed LF display. We established a relati
By pre-adjusting the input elementary images using calibrated coefficients, the impact of these aberrations was compensated
Simulation results, using a practical LF model, indicated improved accuracy of focusing effects and an elevated observed Spati
Abstract: The conflicting visual cues, specifically, the vergence-accommodation conflict (VAC), constitute one of the most signi
of accommodation-invariance. The analysis conducted in comparison with the existing stereo displays and the more advanced
introducing minimal increase in the hardware and software complexities of traditional stereo displays.
Abstract: Reconstruction of defects in the oral cavity and all levels of the pharynx are technically challenging. Free tissue trans
(OrbEye system, Olympus Germany, Hamburg) when performing microvascular anastomosis of the freely transferred radial fo
Abstract: Stereoscopic 3D remote vision system (sRVS) design can be challenging. The components often interact such that ch
elements of the sRVS design. In this study, we investigated the trade-offs between two parameters: viewing distance and cam
measures of visual stress. Conversely, a small amount of camera toe-in had no effect on user performance or comfort. These r
Abstract: TBA
Abstract: During natural viewing, the oculomotor system interacts with depth information through a correlated, tightly related
introduced in a stereoscopic remote vision system (sRVS), the individual elements of the near response may decouple (e.g., ve
potentially preserve image quality in the presence of significant distortion. In this experiment, we measured two elements of t
compression of the image space, and increasing the VA mismatch. We found a strong positive cross-correlation of vergence po
maintain image quality and increase the depth of focus through pupil miosis.
Abstract: We present a robust and reliable method for automatic identification of key features of scanned stereographs which
images), a key points selection step (homologous points in the left and right images) and an image correction and disparity cal
Abstract: Stereoscopic conversion of feature films , like any post-production work, can be carried out with different quality crit
CG films seems to have a simple native solution by adding a stereo camera rig, this also have it’s drawbacks and weaknesses.
In our experience, a single person could achieve a cost-effective stereo conversion , even surpassing the stereoscopic quality o
Here we show how we did it, showing the pros and cons of this depthmap based method , as a cost-effective stereoscopic m
Abstract: TBA
Abstract: This study evaluates user experiences in Virtual Reality (VR) and Mixed Reality (MR) systems during task-based intera
demonstrated significant differences among conditions, with MR (Physical) consistently outperforming MR (CG) and VR in vari
conditions with a higher proportion of CG content, users spent more time observing objects but reported increased discomfor
improving user experiences in VR and MR systems, with potential applications across various domains.
Abstract: Current gold standard for clinical training is on actual Linear Accelerators (LINAC) machines, recordings of the actual
always represent actual use case scenarios. Immersive experience by interacting with LINAC machines in a safe and controlled
Current real world radiation treatment training often requires radiation bunker and actual equipment that is not always acces
Accelerator (LIANC) due to its multi-million dollar capital cost makes the accessibility even worse. To solve the logistical, econo
anywhere and anytime.
We are introducing a virtual reality (VR) space for Radiation trainees, so that they can use it anywhere and anytime to practice
Abstract: Empathy and Perspective shift is traditionally used in design as a means to gain insights about users. Arguably, empa
In August 2023, a co-design workshop was run in a collaboration between Kyushu University (Fukuoka, Japan) and Royal Colle
Each set of drawings demonstrates a specific interaction each participant had with an avian species in three different contexts
? the interaction from the participant’s perspective,
? the interaction from the bird’s perspective, and
? how the interaction will be embodied in 40 years’ time .
The main purpose of this exercise is to co-imagine a utopian future with more positive interspecies relations between humans
Based on these drawings, we are creating a number of visualizations presenting those perspectives in Virtual Reality.
In this way, we allow viewers to visualise participants' perspective shift through a series of drawings made in response to given
Abstract: This presentation is a hypothetical investigation, prompted by the discovery of a plastic collectible Toy Story Alie
3D using red-cyan anaglyph glasses? Probably not, for reasons described in the presentation, but a possible solution is provide
other types of 3-eyed 3D glasses such as passive polarised and active 3D glasses. Are there any useful insights from this presen
Abstract: Virtual and augmented reality technologies have significantly advanced and come down in price during the last few y
reality laboratory can be quite involved, particularly with large-screen display systems. Thus, this keynote presentation will ou
package. Examples for visualizations and their applications using this environment will be discussed from a variety of discipline
Abstract: Similarity detection seeks to identify similar, but distinct items over multivariate datasets. Often, similarity cannot be
effectiveness of common visual encodings for multivariate data in the context of visual similarity detection. We conducted a u
significant differences in performance between encodings, especially as the number of items increases. Surprisingly, we found
contrast to existing guidelines, we found that filled star plots (Kiviats) outperformed other encodings in terms of scalability and
Abstract: User experiments are essential for informing researchers what an audience is seeing in a chart. User experiments are
data with repeated measurements is the use of (Generalized) Linear Mixed Effects Models (GLME). These models increase the
analyzing results from user experiments is commonly used. We compare the two approaches with example data from psychop
practices of open science and reproducible research by providing open access to all of our code and data.
Abstract: Numerous studies of social media analytics (SMA) shed light upon interesting insights into the information flow in so
transparency and understanding of social media. Among several challenges of SMA, this paper focuses on two issues of 1) inva
persona profiling using generalizing contextual analysis via Natural Language Processing (NLP) technologies could address the
Abstract: Large Language Models (LLMs) have demonstrated a huge impact on education and literacy in recent years. We eval
by two popular LLMs (OpenAI's ChatGPT and Google's Bard) to educate novices on the topic of Parallel Coordinate Plots (PCPs
on the analysis of the expert evaluation, we found that while both the LLMs provided some relevant and practical recommend
completely inapplicable to Parallel Coordinate Plots literacy.
Abstract: This study delves into the domain of data visualization within mobile media planning applications and investigates us
managed and displayed together. Consequently, ensuring the quality of experience for user interaction is challenging mainly in
required for personalized user interaction with media planning data. Specifically, three new modules are conceived, specified
anonymized real-life media planning datasets. They show the following beneficial features are granted: mobile data visualizati
regardless of the combination of technologies used to obtain the visualization solution.
Abstract: Web visualization dashboards are popular. We propose a system called RAIV that can capture and archive web visua
experience without needing to install any additional software. RAIV supports intelligent search as well. When a search target h
and publicly available census data visualizations from US Census.
Abstract: We present an interactive visualization tool to explore high-dimensional features of audiovisual data extracted from
features.
Features are extracted using neural networks, signal processing techniques, and audio analysis tools. Furthermore, we presen
We illustrate our approach via use cases showing initial results to analyze song features, compare songs, identify outstanding
Abstract: The rise in crime rates over the past few years is a major issue and is a huge source of worry for police departments
large cities often see criminal activity. Many studies, media, and websites include statistics on crime and it is contributing elem
Denton. We assess areas that are significantly affected based on zip codes and variations in crime categories. As the crime rate
show crime frequency and distribution across four different cities and supply valuable information about the complex relation
incidents onto a geographical map, and supply insight into factors affecting crime that will help them deploy resources and he

EI2024 Paper-Abstract Searchable

Uploaded by

Copyright:

Available Formats

EI2024 Paper-Abstract Searchable

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

EI2024 Paper-Abstract Searchable

Uploaded by

Copyright:

Available Formats

Paper

TITLE First Author Conference

JIST-first: Camera Motion Estimation

A VGA Light-Tracing Time-of-Flight CMOS

Extending Lidar Depth Range Using

A Deep Learning Based Light Field Image

Data Augmentation Based on Depth Hamed Razavi

N-DEPTH: Neural Depth Encoding for Stephen

Evaluation of Signal and Noise Metrics of

Measurement of Noise Equivalent

An Integrated and Fully Automated Test

Multi-Modal Pedestrian Detection via

Exploring Object Detection and Image

Efficient Fault Tolerant Architecture for Shyam

Fusing Segmentation and Domain-

Optimizing Ego Vehicle Trajectory

An Approach to Recreate Large Virtual

Urban Scene Reconstruction based on

Provable Probabilistic Imaging Using

Transformers for Microscopy Slide Image

Generative Plug-and-Play: Posterior

Keynote: Efficient Neural Scene

COIMG-122 Dark Secrets of Computational Imaging Laura Waller COIMG

Super Resolution in Heterogeneous Array

Diffractive Processing of Visual

Physics-informed Generative Models for Muhammad

Considerations for Materials Design from

Methodology for Estimation of Intrinsic

Persistent Homology for Microstructure

Mathematical Nuances of Gaussian-

Reinforcement Learning for Metals

Evaluation of Information Content in 2D

Semantic Segmentation and Anomaly

COIMG Keynote: Quantitative Secondary

Biomedical Ultrasound Computed

Automated, High-Throughput X-ray CT

COIMG KEYNOTE: Understanding

Multimodal Registration and Fusion of In

Benefit of X-ray Imaging to Improve Anne-Françoise

Model-based Deep Learning with a CNN

Leveraging Pixel Value Certainty in Pixel-

Multimodal Deep Learning Approach for

Gradient-Based Image Visualization for

COIMG KEYNOTE: Breaking Boundaries in

Generative Denoising Models for Patxi Fernandez-

Classification of Melt Pool Boundaries

JIST-first : Computational Image

Beyond RISC - Next-generation Compute

COIMG-152 PSF Engineering for Edge Sensing Nicholas Antipa COIMG

Reconstruction of Phase and

COIMG-155 Mobile Aware Computational Imaging Jing Li COIMG

Accurate Efficient Multispectral

Neural Deconvolution and Rendering for

Implicit Neural Representation for

Maximum Likelihood-based Phase-

Detection of Multiple Fluorescent Dyes

Open Set Domain Adaptation for Image

3D Printing Images Defect Detection and

3D Printing Images Defect Detection and

Experimental Study on Reducing