Object Tracking (Computer Vision)

Surveillance with facial recognition holds immense potential as a technological tool for combating crime in Latin American countries. However, the limitations of fixed cameras in covering wide areas and tracking suspects as the evaded recognitions systems pose significant challenges. To address these limitations, we propose a facial recognition system designed to recognize faces of suspected individuals with criminal backgrounds and missing persons. Our solution combines facial recognition technology with a custom-built unmanned aerial vehicle (UAV) for the identification and tracking of targeted persons listed in a database for crimes. We utilize the inception v2 model to deploy a Siamese network on the Jetson TX2 platform for facial recognition. Additionally, we introduce a novel tracking algorithm to track suspected individuals in the event of evasion. During field test experiments, our system demonstrated strong performance in facial recognition across three different environments: stationary, indoor flight, and outdoor flight. The accuracy of our system was 94.45% for recognizing along with our tracking algorithms. An improvement of 1.5% in recognition and better tracking approach for surveillance. This indicates the versatility and effectiveness of our solution in various operational scenarios, enhancing its potential for crime prevention and law enforcement efforts in Latin American countries. INDEX TERMS Unmanned autonomous vehicles (UAV), facial recognition, object tracking, deep learning, autonomous flight, embedded systems.

Pedestrian tracking and detection have become critical aspects of advanced driver assistance systems (ADASs), due to their academic and commercial potential. Their objective is to locate various pedestrians in videos and assign them unique identities. The data association task is problematic, particularly when dealing with inter-pedestrian occlusion. This occurs when multiple pedestrians cross paths or move too close together, making it difficult for the system to identify and track individual pedestrians. Inaccurate tracking can lead to false alarms, missed detections, and incorrect decisions. To overcome this challenge, our paper focuses on improving data association in our pedestrian detection system’s Deep-SORT tracking algorithm, which is solved as a linear optimization problem using a newly generated cost matrix. We introduce a set of new data association cost matrices that rely on metrics such as intersections, distances, and bounding boxes. To evaluate trackers in real time,...

We present a novyl algorithm for the robust and reliable automatic extraction of lip featu;re points for speechreading. The algorithm uses a combination of colour information in the image data and knowledge about the structure of the mouth area to find certain feature points on the inner lip contour. A new confidence measure quantifying how well the feature extraction process worked is introduced. A parameter set describing the shape of the mouth is. derived from the positions of the feature points. Using a stereo camera system, measurements are in 3D. Such a 3D parameter set is of great value for automatic speech-reading systems.

Resumo Este artigo revisa os avanços no desenvolvimento do Núcleo Avançado de Visualização (NAV). Um laboratório móvel voltado para pesquisa e desenvolvimento de Realidade Aumentada (RA) em ambientes externos e gestão colaborativa. O NAV é composto de uma série de soluções integradas que combinadas visam melhorar o desempenho durante a gestão de projetos, aprimorando o acesso a informações em campo e supervisão remota de projetos. A abordagem utiliza várias estações de trabalho móveis interligadas por uma rede sem fio, que pode fornecer informações baseadas na posição do usuário, tais como, arquivos CAD comentáveis, modelos 3D, visualização através de Realidade Aumentada ao ar livre e constante posicionamento e rastreamento GPS. O sistema também permite supervisão/inspeção por telepresença panorâmica. O trabalho apresenta o conceito geral NAV descrevendo as principais soluções e seu status atual de desenvolvimento e por último, discute a integração da ferramenta e possíveis pesquisa...

Correlation filter plays a major role in improved tracking performance compared to existing trackers. The tracker uses the adaptive correlation response to predict the location of the target. Many varieties of correlation trackers were proposed recently with high accuracy and frame rates. The paper proposes a method to select a single background patch to have a better tracking performance. The paper also contributes a variant of correlation filter by modifying the filter with image restoration filters. The approach is validated using Object Tracking Benchmark sequences.

Figure 1. Creepy Tracker is an open-source toolkit that provides spatial information about people and interactive surfaces. To do this, it resorts to multiple depth sensing cameras (A, B). It helps the design of systems that handle, for instance, (C) interactive tabletops, (D) vertical surfaces, (E) floor projections and even capture avatars for (F) telepresence or (G) virtual reality.

In this paper, a new 3D chaotic system is controlled by generalized backstepping method. Generalized backstepping method is similarity to backstepping method but generalized backstepping method is more applications in systems than it. Backstepping method is used only to strictly feedback systems but generalized backsteppingmethod expand this class. New 3D chaotic system is controlled in two participate sections; stabilization and tracking reference input. Numerical simulations are presented to demonstrate the effectiveness of the controlschemes.

Régression sous contraintes : application à l'estimation de la courbe granulométrique d'un aérosol Revue de statistique appliquée, tome 19, n o 4 (1971), p. 23-44 <http://www.numdam.org/item?id=RSA_1971__19_4_23_0> © Société française de statistique, 1971, tous droits réservés. L'accès aux archives de la revue « Revue de statistique appliquée » (http://www. sfds.asso.fr/publicat/rsa.htm) implique l'accord avec les conditions générales d'utilisation (http://www.numdam.org/conditions). Toute utilisation commerciale ou impression systématique est constitutive d'une infraction pénale. Toute copie ou impression de ce fichier doit contenir la présente mention de copyright. Article numérisé dans le cadre du programme Numérisation de documents anciens mathématiques http://www.numdam.org/

Os profissionais, pesquisadores, educadores e estudantes com atuação e interesse na área de Realidade Virtual (RV), e mais recentemente Realidade Aumentada (RA), vêm crescendo em número e qualidade. Prova disso é o número crescente de participantes a cada nova edição do Symposium on Virtual Reality (SVR), no qual tem-se presenciado aumento constante na qualidade e quantidade dos artigos submetidos. Além disso, percebe-se o surgimento contínuo de novos cursos, empresas e grupos de pesquisa em tecnologia de RV e RA e suas aplicações. Somos uma comunidade jovem, multidisciplinar, criativa, aberta a idéias e desafios, a qual atrai e acolhe novos talentos, sejam eles estudantes ou profissionais migrando de outras áreas. Foi pensando nesses novos e bem-vindos participantes do SVR que criamos o Pré-Simpósio (PS), apresentado pela primeira vez em São Paulo junto ao SVR 2004. O sucesso da iniciativa fez com que a Comissão Especial de Realidade Virtual da SBC, responsável pela organização e promoção do SVR, incluísse de forma definitiva o PS na programação de atividades do evento. O principal objetivo do PS é oferecer um curso rápido e abrangente sobre os principais conceitos e tecnologias das áreas de RV e RA, de tal forma a estabelecer um repertório básico que ajude o participante a melhor aproveitar tudo o que será exibido e discutido ao longo dos três dias de atividades principais do SVR. Criado, desenvolvido e apresentado por professores e pesquisadores seniores da comunidade de RV e RA, o Pré-Simpósio oferece aos participantes, além das 8 horas-aula, material complementar na forma de um texto abrangente que cobre os principais conceitos e tecnologias da área, incluindo um glossário, cujo conteúdo vai muito além do que é apresentado ao vivo. No SVR 2004 o PS deu origem ao livro "Realidade Virtual: Conceitos e Tecnologia". Esse livro, já esgotado, tem sido usado como referência em cursos técnicos e superiores, não só da área de computação e informática, mas também de design, comunicação e artes. Para o SVR 2006 procedemos a uma reestruturação e revisão da publicação do Pré-Simpósio do SVR 2004, com a revisão e ampliação de capítulos e criação de novos. Esse trabalho deu origem a este livro, denominado "Fundamentos e Tecnologia de Realidade Virtual e Aumentada". Espera-se que este livro venha contribuir para a disseminação e consolidação da Realidade Virtual nos meios acadêmicos e profissionais. Portanto é com grande prazer que apresentamos este resultado do trabalho de 56 autores da comunidade brasileira de RV e um autor internacional, que não mediram esforços para produzir este texto didático e de qualidade. Desejamos que este o ajude a ingressar nesse fascinante mundo, para não só aproveitá-lo como também contribuir para sua evolução (quem sabe você não estará escrevendo um dos capítulos desta publicação no futuro?). Seja bem-vindo e tenha uma boa "viagem".

UAVs have been deployed in various object tracking applications such as disaster management, traffic monitoring, wildlife monitoring and crowd management. Recently, various deep learning methodologies have a profound effect on object detection and tracking. Deep learning-based object detectors rely on pre-trained networks. Problems arise when there is a mismatch between the pre-trained network domain and the target domain. UAV images possess different characteristics than images used in pre-trained networks due to camera view variation, altitude ranges and camera motion. In this paper, we propose a novel methodology to detect and track objects from UAV data. A deeply supervised object detector (DSOD) is entirely trained on UAV images. Deep supervision and dense layer-wise connection enriches the learning of DSOD and performs better object detection than pre-trained-based detectors. Long-Short-Term Memory (LSTM) is used for tracking the detected object. LSTM remembers the inputs from the past and predicts the object in the next frame thereby bridging the gap of undetected objects which improves tracking. The proposed methodology is compared with pre-trained-based models and it outperforms.

The application of Transformers in computer vision has gained momentum, even to the extent of revising the Vision Transformer (ViT) theory of abandoning CNNs, or to be exact CNN backbones for Transformer-based backbones. This research attempts to evaluate the efficiency backbones when incorporated into a re-ID-based model such as FairMOT which is traditionally trained using a CNN. We investigate how Transformer-based feature extraction impacts tracking performance, particularly for small and occluded objects such as fish in video data. Our findings indicate that while ViT backbones offer promising features, they do not yet surpass CNN-based methods in terms of tracking accuracy in regards to the FairMOT approach. This study highlights the need for further optimization of Transformer architectures.

Feature Selection Library (FSLib) is a widely applicable MATLAB library for Feature Selection (FS). FS is an essential component of machine learning and data mining which has been studied for many years under many different conditions and in diverse scenarios. These algorithms aim at ranking and selecting a subset of relevant features according to their degrees of relevance, preference, or importance as defined in a specific application. Because feature selection can reduce the amount of features used for training classification models, it alleviates the effect of the curse of dimensionality, speeds up the learning process, improves model's performance, and enhances data understanding. This short report provides an overview of the feature selection algorithms included in the FSLib MATLAB toolbox among: filter, embedded, and wrappers methods.

In an era where accumulating data is easy and storing it inexpensive, feature selection plays a central role in helping to reduce the high-dimensionality of huge amounts of otherwise meaningless data. This report overviews concepts and algorithms of feature selection, surveys existing feature selection algorithms for classification with a categorizing framework based on the complexity: filter, embedded, and wrappers methods. Some real-world applications are included to demonstrate the use of feature selection in data mining. As a result, the report proposes extensive tests on diverse datasets. We conclude this work by identifying trends and challenges of feature selection research and development while providing a toolbox of 10 methods selected from recent literature.

In an era where accumulating data is easy and storing it inexpensive, feature selection plays a central role in helping to reduce the high-dimensionality of huge amounts of otherwise meaningless data. This manuscript overviews concepts and algorithms of feature selection, surveys existing feature selection algorithms for classification based on the complexity: filter, embedded, and wrappers methods. Some real-world applications are included. We conclude this work by identifying trends and challenges of feature selection research and development while providing a code library of methods selected from recent literature.

Video segmentation plays an important role in computer vision. The process of video segmentation uses in various fields such as road transportation, security surveillance and medical diagnosis of critical stage of human body damage detection. Various authors proposed various methods for the segmentation of video such as color based segmentation, texture based segmentation and shape based segmentation. In this paper modified the HEM clustering technique for video segmentation using genetic algorithm. Genetic algorithm is population based optimization technique. The genetic algorithm used for the selection of M estimation parameter. The value of M decides the better clustering process. For the empirical evaluation used MATLAB software and Google texture video data and YouTube video data. The proposed method improved the value of F-measure and accuracy in compression of HEM method. The proposed clustering algorithm is capable of both clustering Dynamic texture and searching novel Dynam...

Now a day's vast growth of our technological world and infrastructure has made our lives easiest manner in existing system. The advent of science and Technology has also increased the traffic prone hazards analyzing using Traffic Management System and avoid the road accidents take place frequently which rapidly causes represents huge loss of our life's and property because of the lowest emergency facilities. The vehicle detection System will provide detected and an optimum solution to this drawback to overcome on existing environment. An accelerometer can be used in a vehicle detection application so that dangerous operations on driving can be detected. It can be used as a crash or rollover detector of the vehicle during and after a crash. With signals from an ultrasonic sensor, a severe accident due to an obstacle can be recognized. When a vehicle meets with an accident or if a vehicle rolls over, the accelerometer and ultrasonic sensor automatically detects the signal processing and immediately sends it to microcontroller then immediate response. Microcontroller sends the detected alert message through required the GSM module including the location to existing Government or Private bodies control room or a rescue team. So the emergency help team can immediately trace the location through the GPS module, after receiving the information from existing resources. The required location can also be viewed through Google maps and ISRO Bhuvan. After conforming the location necessary action is taken through necessary bodies. If the person meets with a small accident or if there is no serious threat to anyone`s life, then the alert message can be terminated by the driver by a switch provided in order to avoid wasting the valuable time of the medical rescue team, if required operation is necessary then alerted through PDA device on installed apps. We proposed these work is useful to identify Traffic load analyzing and vehicle tracking and detecting when the accident precisely occurred by means of both accelerometer and ultrasonic sensor combine working during Transportation using Internet of Things applications.

Use a deep learning-based computer vision method for vehicle detection and tracking, plays an integral role in intelligent transportation systems and traffic accident detection. Road Vehicle detection, recognition, and counting are an important part of road traffic monitoring and surveillance. Designing a road monitoring model with good performance is very complex. Traditional vehicle detection capabilities based on artificial intelligence algorithms have poor detection capabilities and robustness. This paper proposes an extended efficient single stage detection technique that is based on the YOLO to perform vehicle detection and multi-object tracking methods for tracking and counting vehicles that are based on a computer vision approach. The proposed deep learning model takes a moving vehicle's video, proposed detection technique vehicles detect and create a unique id for each of the initial detection after then tracking each of the vehicles as they move around frames in a given video, maintaining the id assignment. The proposed detection algorithm has great adaptability when viewed under a variety of conditions, such as heavy traffic, night environment, multiple vehicles overlapping, and part of a vehicle missing. Experimental results show that the algorithm can accurately detect and recognize vehicles according to their edge contours.

The difficulties that all traffic monitoring or road surveillance team encounters in the regulation and management of smart roads are prominent research areas for researchers or scientists. Algorithms for computer vision and machine learning are frequently used to analyse vast amounts of gathered video or picture data. For the purpose of detecting and tracking vehicles from video datasets, in this paper, we propose an efficient deep learning approach to computer vision. By applying the effective optimization technique framework to train a one-stage detector, we attempt Single-stage systems are used in the proposed deep learning methods to increase the precision of vehicle detection and the speed at which results may be computed. Single-stage detectors improve testing speed while two-stage trackers have great local availability and accurate object recognition. To determine trajectories, the Deep-SORT tracker method is employed to find bonding boxes. We analyse and test certain characteristics like Precision-Recall and Mean Average Precision in the highway traffic video dataset from Kaggle to determine how well the proposed Deep Learning technique performs.

s a fundamental task for research in satellite videos (SVs), object tracking is used to track the target of interest in traffic evaluation, military security, and so forth. The current satellite technology in the remote sensing field makes it possible to track moving targets with a relatively high frame rate and image resolution. However, objects under this special view are often small and blurry, making it hard to extract deep features effectively. As a result, quite a few deep learning (DL) methods were proposed for object tracking in SVs. In addition, evaluation criteria for daily life videos (DLVs) are not fully applicable to SVs, which always get low precision evaluation results for tiny objects. In this article, we make three contributions to the research on SVs. First, a new single object tracking (SOT) dataset, named SV248S, is proposed, including 248 sequences with high-precision manual annotation, and 10 kinds of attribute tags are designed to completely represent the difficulties during tracking. Second, two highprecision evaluation methods are proposed, especially for small object tracking. Finally, 28 DL-based state-of-the-art (SOTA) tracking methods, from 2017 to 2021, covering popular frameworks, are evaluated and compared on the proposed dataset. Furthermore, some guidelines for effectively adopting DL-based methods are summarized based on comprehensive experimental results. BACKGROUND ON TRACKING METHODS Tracking objects in SVs has been a valuable research tool over the years. The dynamic scene carries much more spatial and temporal information than the static image. Although DL-based methods have been successfully implemented in many tasks in the remote sensing field [1], there are still fewer DL-based tracking methods for SVs. The SVs captured by very-high-resolution (VHR) remote sensing technology have been used in many applications (which are introduced in [2], [3]), such as traffic evaluation; geological and natural disaster monitoring; and military security. The task of SOT

The objective of this paper is to early detect security camera tampering with the proposed algorithm to address the two major issues of camera tampering attacks such as camera defocus and camera-lens block. These anomaly events such as screen shaking, flickering, color casting and lens cover significantly impact the effectiveness of video surveillance systems. The purpose of the algorithm is to measure the distance between a moving object and the camera's field of view. Because of this close distance estimation can detect camera tampering earlier. To evaluate the performance of the proposed algorithm, three video datasets with various types of camera attacks were tested with 64 anomalous events in video sequences. The proposed algorithm demonstrated its efficiency with an average of 3.1% missing events and an average of 7.8% false alarms in the experimental results.

UAVs have been deployed in various object tracking applications such as disaster management, traffic monitoring, wildlife monitoring and crowd management. Recently, various deep learning methodologies have a profound effect on object detection and tracking. Deep learning-based object detectors rely on pre-trained networks. Problems arise when there is a mismatch between the pre-trained network domain and the target domain. UAV images possess different characteristics than images used in pre-trained networks due to camera view variation, altitude ranges and camera motion. In this paper, we propose a novel methodology to detect and track objects from UAV data. A deeply supervised object detector (DSOD) is entirely trained on UAV images. Deep supervision and dense layer-wise connection enriches the learning of DSOD and performs better object detection than pre-trained-based detectors. Long-Short-Term Memory (LSTM) is used for tracking the detected object. LSTM remembers the inputs from the past and predicts the object in the next frame thereby bridging the gap of undetected objects which improves tracking. The proposed methodology is compared with pre-trained-based models and it outperforms.

UAVs have been deployed in various object tracking applications such as disaster management, traffic monitoring, wildlife monitoring and crowd management. Recently, various deep learning methodologies have a profound effect on object detection and tracking. Deep learning-based object detectors rely on pre-trained networks. Problems arise when there is a mismatch between the pre-trained network domain and the target domain. UAV images possess different characteristics than images used in pre-trained networks due to camera view variation, altitude ranges and camera motion. In this paper, we propose a novel methodology to detect and track objects from UAV data. A deeply supervised object detector (DSOD) is entirely trained on UAV images. Deep supervision and dense layer-wise connection enriches the learning of DSOD and performs better object detection than pre-trained-based detectors. Long-Short-Term Memory (LSTM) is used for tracking the detected object. LSTM remembers the inputs from the past and predicts the object in the next frame thereby bridging the gap of undetected objects which improves tracking. The proposed methodology is compared with pre-trained-based models and it outperforms.

The capability of the unmanned aerial vehicle (UAV) to capture highly informative data has expanded its utility in multiple sectors. Surveillance-based UAV applications have highly relied on accurate object tracking. During UAV monitoring, issues such as changes in object appearance and occlusion are common. Tracking the objects under such a scenario is a challenging task. On the other hand, accurate object tracking is quintessential in critical scenarios like security surveillance. In this work, a novel deep learning-based framework for accurate multiple objects tracking with UAV videos is proposed. Tiny-Deeply Supervised Object Detector (Tiny-DSOD) is adopted for accurate object detection. A novel stacked bidirectional-forward LSTM (SBF-LSTM) tracker with spatial and visual features is proposed for object tracking. The spatial and visual features obtained from Tiny-DSOD are trained with the tracker, which predicts object location during tracking. The choice of SBF-LSTM as the tracker enables accurate prediction of object location. Object association is dealt with based on bounding box distance, appearance, and size metrics. With the proposed model, switching object identities is lessened to a greater extent, thereby increasing the tracking accuracy. The proposed methodology outperforms the state-ofthe-art methods on UAV videos.

In this paper a proposal for the tracking moving objects and their movements description taken from sequences of images is presented. Some of the properties that can be inferred for each moving object are the magnitude and direction of motion. In the first part, the segmentation of moving objects is presented. This segmentation is made from optical flow without requiring a priori information of the scene. The estimation of optical flow is obtained by Pyramid Lucas-Kanade algorithm. Optical flow vectors are grouped by proximity, direction and magnitude. The segmentation of the objects in regions is made from the convex hull of origins of each optical flow vector wich forms the region. In the second part, the velocity and direction of movement of each region is obtained.

Searching test image from image databases using features extraction from the content is currently an active research area. In this work, we present novel feature extraction approaches for content-based image retrieval, when the query image is color image. To facilitate robust man-machine interfaces, we accept query images with color attributes. Special attention is given to the similarity measure with different distance matrices properties since the test image and object image from database finding the distance measuring. Several applicable techniques within the literature are studied for these conditions. The goal of this paper is to present the user with a subset of images that are more similar to the object image. One of the most important aspects of the proposed methods is the accuracy measurement of the query image with different database images. The method significantly improves the feature extraction process and enables it, to be used for other computer vision applications.

In application of tracking and detecting the suspicious activities, multiple object tracking (MOT) has been given fine attention due to its application as it provides the parallel task of identification and tracking of human. MOT ensures the identification and trajectory for each object frame as they interact, despite the changes in its appearance, occlusion and various other tasks involved. Recent adoption of deep learning has given a new perspective but still achieving high metrics remains a major issue to overcome such issues, this research work presents the integrated architecture of deep convolutional covariance networks (DCCNs) and space-time adaptive correlation tracking (STACT) algorithm with similarity map function (SMF). Moreover, in proposed work, DCCNs is utilized for feature extractions through each frame capturing the distinctive information, STACT is tracking approaches that utilizes the SMF for locating and tracking objects. SMFs are updated for any changes in human appearances and motion, also it deals with occlusion. Here the proposed model is evaluated on MOT17 and MOT20 dataset. Performance analysis is carried out through comparing the existing model and Integrated-DCCN achieves higher metrics.

This paper presents experimental insights from the deployment of an ensemble of heterogeneous autonomous sensor systems over a shallow coral reef. Visual, inertial, GPS, and ultrasonic data collected are compared and correlated to produce a comprehensive view of the health of the coral reef. Coverage strategies are discussed with a focus on the use of informed decisions to maximize the information collected during a fixed period of time.

This article presents an adaptive Bayes model for the decision logic of cascade classifier structures. The proposed method is fast and robust with respect to multimodal and overlapping distributions and can be applied to arbitrary stage classifiers with continuous outputs. The method consists of an adaptive computation of thresholds and probability density functions which outperform the threshold based decision. It furthermore guarantees high detection rates independent of the number of stage classifiers. Based on this Bayes model different confidence measures are proposed and evaluated statistically and used for merging detection windows. The algorithm is applied to the detection of US speed limit signs under typical driving conditions. Results show that on a single CPU with 3.3 GHz the proposed method yields single image detection rates of 97 % with 0.2 false positives per image running at 13 Hz, and for a different setup a detection rate of 93 % with 0.2 false positives per image performing with 43 Hz for scanning the whole image (752x480 pixels).

In the circumstance of the COVID-19 pandemic, Prevention is better than cure, especially if the cure is not available, the first motto that all health organizations recommend is keep distances between people to prevent epidemic spread. In this paper, an online multi layers social distance detection system is proposed, the main idea is to detect distance among pupils and classify the distance to accept or not, this system treats stream video of fixed camera which monitor the whole school yard where the pupils are available, this proposed system used multi layers, the first is to make person detection using Yolo-4 approach including CNN model, and surround it by rectangle, the second is to specify the center of detected person, finally, calculate the relative distance to decide if it is accepted or not, this system works online and give high accuracy.

One of the most difficult perceptual problems for many applications is accurately recognizing the human object in a variety of circumstances. This can be difficult due to obstructions, weather, complex backdrops, cast shadows, and occlusions. Occlusion is a challenging open problem where a detector can only perceive a portion of the target human because of obstacles in the surrounding. In this research, an experimental investigation was conducted using the multi object tracking (MOT17) datasets to construct a graph neural network-based solution for the detection of humans in videos while considering the possibility of occlusion. Graph neural network (GNN) is used for the construction of neural solver model for detecting human object in occlusion scenario. The results obtained shows that this proposed method offers a considerable improvement in efficiency in comparison to the ways that have been used in the past. The values obtained for the standard performance metrics are higher than the state-of-the-art methods.

This paper proposes a new approach to multi-object tracking by semantic topic discovery. We dynamically cluster frame-by-frame detections and treat objects as topics, allowing the application of the Dirichlet process mixture model. The tracking problem is cast as a topic-discovery task, where the video sequence is treated analogously to a document. It addresses tracking issues such as object exclusivity constraints as well as tracking management without the need for heuristic thresholds. Variation of object appearance is modeled as the dynamics of word co-occurrence and handled by updating the cluster parameters across the sequence in the dynamical clustering procedure. We develop two kinds of visual representation based on super-pixel and deformable part model and integrate them into the model of automatic topic discovery for tracking rigid and non-rigid objects, respectively. In experiments on public data sets, we demonstrate the effectiveness of the proposed algorithm.

This paper proposes a new approach to multi-object tracking by semantic topic discovery. We dynamically cluster frame-by-frame detections and treat objects as topics, allowing the application of the Dirichlet Process Mixture Model (DPMM). The tracking problem is cast as a topic-discovery task where the video sequence is treated analogously to a document. This formulation addresses tracking issues such as object exclusivity constraints as well as cannot-link constraints which are integrated without the need for heuristic thresholds. The video is temporally segmented into epochs to model the dynamics of word (superpixel) co-occurrences and to model the temporal damping effect. In experiments on public data sets we demonstrate the effectiveness of the proposed algorithm.

This paper proposes a new approach to multi-object tracking by semantic topic discovery. We dynamically cluster frame-by-frame detections and treat objects as topics, allowing the application of the Dirichlet process mixture model. The tracking problem is cast as a topic-discovery task, where the video sequence is treated analogously to a document. It addresses tracking issues such as object exclusivity constraints as well as tracking management without the need for heuristic thresholds. Variation of object appearance is modeled as the dynamics of word co-occurrence and handled by updating the cluster parameters across the sequence in the dynamical clustering procedure. We develop two kinds of visual representation based on super-pixel and deformable part model and integrate them into the model of automatic topic discovery for tracking rigid and non-rigid objects, respectively. In experiments on public data sets, we demonstrate the effectiveness of the proposed algorithm.

Micro expression recognition (MER)is a very challenging task as the expression lives very short in nature and demands feature modeling with the involvement of both spatial and temporal dynamics. Existing MER systems exploit CNN networks to spot the significant features of minor muscle movements and subtle changes. However, existing networks fail to establish a relationship between spatial features of facial appearance and temporal variations of facial dynamics. Thus, these networks were not able to effectively capture minute variations and subtle changes in expressive regions. To address these issues, we introduce an active imaging concept to segregate active changes in expressive regions of a video into a single frame while preserving facial appearance information. Moreover, we propose a shallow CNN network: hybrid local receptive field based augmented learning network (OrigiNet) that efficiently learns significant features of the micro-expressions in a video. In this paper, we propose a new refined rectified linear unit (RReLU), which overcome the problem of vanishing gradient and dying ReLU. RReLU extends the range of derivatives as compared to existing activation functions. The RReLU not only injects a nonlinearity but also captures the true edges by imposing additive and multiplicative property. Furthermore, we present an augmented feature learning block to improve the learning capabilities of the network by embedding two parallel fully connected layers. The performance of proposed OrigiNet is evaluated by conducting leave one subject out experiments on four comprehensive ME datasets. The experimental results demonstrate that OrigiNet outperformed state-of-the-art techniques with less computational complexity.

Unlike prevalent facial expressions, micro expressions have subtle, involuntary muscle movements which are short-lived in nature. These minute muscle movements reflect true emotions of a person. Due to the short duration and low intensity, these micro-expressions are very difficult to perceive and interpret correctly. In this paper, we propose the dynamic representation of micro-expressions to preserve facial movement information of a video in a single frame. We also propose a Lateral Accretive Hybrid Network (LEARNet) to capture micro-level features of an expression in the facial region. The LEARNet refines the salient expression features in accretive manner by incorporating accretion layers (AL) in the network. The response of the AL holds the hybrid feature maps generated by prior laterally connected convolution layers. Moreover, LEARNet architecture incorporates the cross decoupled relationship between convolution layers which helps in preserving the tiny but influential facial muscle change information. The visual responses of the proposed LEARNet depict the effectiveness of the system by preserving both high-and micro-level edge features of facial expression. The effectiveness of the proposed LEARNet is evaluated on four benchmark datasets: CASME-I, CASME-II, CAS(ME)^2 and SMIC. The experimental results after investigation show a significant improvement of 4.03%, 1.90%, 1.79% and 2.82% as compared with ResNet on CASME-I, CASME-II, CAS(ME)^2 and SMIC datasets respectively.

We present a novel solution to the problem of 3D tracking of the articulated motion of human hand(s), possibly in interaction with other objects. The vast majority of contemporary relevant work capitalizes on depth information provided by RGBD cameras. In this work, we show that accurate and efficient 3D hand tracking is possible, even for the case of RGB stereo. A straightforward approach for solving the problem based on such input would be to first recover depth and then apply a state of the art depth-based 3D hand tracking method. Unfortunately, this does not work well in practice because the stereo-based, dense 3D reconstruction of hands is far less accurate than the one obtained by RGBD cameras. Our approach bypasses 3D reconstruction and follows a completely different route: 3D hand tracking is formulated as an optimization problem whose solution is the hand configuration that maximizes the color consistency between the two views of the hand. We demonstrate the applicability of our method for real time tracking of a single hand, of a hand manipulating an object and of two interacting hands. The method has been evaluated quantitatively on standard datasets and in comparison to relevant, state of the art RGBD-based approaches. The obtained results demonstrate that the proposed stereo-based method performs equally well to its RGBD-based competitors, and in some cases, it even outperforms them.

The Visual Object Tracking challenge VOT2018 is the sixth annual tracker benchmarking activity organized by the VOT initiative. Results of over eighty trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in the recent years. The evaluation included the standard VOT and other popular methodologies for short-term tracking analysis and a "real-time" experiment simulating a situation where a tracker processes images as if provided by a continuously running sensor. A long-term tracking subchallenge has been introduced to the set of standard VOT sub-challenges. The new subchallenge focuses on long-term tracking properties, namely coping with target disappearance and reappearance. A new dataset has been compiled and a performance evaluation methodology that focuses on long-term tracking capabilities has been adopted. The VOT toolkit has been updated to support both standard short-term and the new longterm tracking subchallenges. Performance of the tested trackers typically by far exceeds standard baselines. The source code for most of the trackers is publicly available from the VOT page. The dataset, the evaluation kit and the results are publicly available at the challenge website 60 .

The usage of both off-the-shelf and end-to-end trained deep networks have significantly improved performance of visual tracking on RGB videos. However, the lack of large labeled datasets hampers the usage of convolutional neural networks for tracking in thermal infrared (TIR) images. Therefore, most state of the art methods on tracking for TIR data are still based on handcrafted features. To address this problem, we propose to use image-to-image translation models. These models allow us to translate the abundantly available labeled RGB data to synthetic TIR data. We explore both the usage of paired and unpaired image translation models for this purpose. These methods provide us with a large labeled dataset of synthetic TIR sequences, on which we can train end-to-end optimal features for tracking. To the best of our knowledge we are the first to train end-to-end features for TIR tracking. We perform extensive experiments on VOT-TIR2017 dataset. We show that a network trained on a large dataset of synthetic TIR data obtains better performance than one trained on the available real TIR data. Combining both data sources leads to further improvement. In addition, when we combine the network with motion features we outperform the state of the art with a relative gain of over 10%, clearly showing the efficiency of using synthetic data to train end-to-end TIR trackers.

Robust visual tracking is a challenging computer vision problem, with many real-world applications. Most existing approaches employ hand-crafted appearance features, such as HOG or Color Names. Recently, deep RGB features extracted from convolutional neural networks have been successfully applied for tracking. Despite their success, these features only capture appearance information. On the other hand, motion cues provide discriminative and complementary information that can improve tracking performance. Contrary to visual tracking, deep motion features have been successfully applied for action recognition and video classification tasks. Typically, the motion features are learned by training a CNN on optical flow images extracted from large amounts of labeled videos. This paper presents an investigation of the impact of deep motion features in a tracking-by-detection framework. We further show that hand-crafted, deep RGB, and deep motion features contain complementary information. To the best of our knowledge, we are the first to propose fusing appearance information with deep motion features for visual tracking. Comprehensive experiments clearly suggest that our fusion approach with deep motion features outperforms standard methods relying on appearance information alone.

The Thermal Infrared Visual Object Tracking challenge 2016, VOT-TIR2016, aims at comparing short-term single-object visual trackers that work on thermal infrared (TIR) sequences and do not apply pre-learned models of object appearance. VOT-TIR2016 is the second benchmark on short-term tracking in TIR sequences. Results of 24 trackers are presented. For each participating tracker, a short description is provided in the appendix. The VOT-TIR2016 challenge is similar to the 2015 challenge, the main difference is the introduction of new, more difficult sequences into the dataset. Furthermore, VOT-TIR2016 evaluation adopted the improvements regarding overlap calculation in VOT2016. Compared to VOT-TIR2015, a significant general improvement of results has been observed, which partly compensate for the more difficult sequences. The dataset, the evaluation kit, as well as the results are publicly available at the challenge website.

Visual object tracking is a classical, but still open research problem in computer vision, with many real world applications. The problem is challenging due to several factors, such as illumination variation, occlusions, camera motion and appearance changes. Such problems can be alleviated by constructing robust, discriminative and computationally efficient visual features. Recently, biologically-inspired channel representations [9] have shown to provide promising results in many applications ranging from autonomous driving to visual tracking. This paper investigates the problem of coloring channel representations for visual tracking. We evaluate two strategies, channel concatenation and channel product, to construct channel coded color representations. The proposed channel coded color representations are generic and can be used beyond tracking. Experiments are performed on 41 challenging benchmark videos. Our experiments clearly suggest that a careful selection of color feature together with an optimal fusion strategy, significantly outperforms the standard luminance based channel representation. Finally, we show promising results compared to state-of-the-art tracking methods in the literature.