327 questions
-1
votes
0
answers
20
views
Should I use YOLO for this purpose? [closed]
I am a beginner in deep learning.
I am doing a project on "Landslide Detection Using Satellite Imagery". I am a bit torn on what to use ?
Should I use YOLO for this purpose or U-Net, or R-...
0
votes
0
answers
20
views
How to extract RoI Features from Faster-RCNN?
I want to use a pretrained Faster-RCNN to extract RoI Features for the detected objects in images.
I couldn't find an easy way of doing that so I started reading the source code. The solution I ...
- 1
0
votes
0
answers
30
views
Training Result and GPU Memory issues when training Faster RCNN model from PyTorch library
Approach
I attempted to train the pretrained fasterrcnn_resnet50_fpn_v2 model from the PyTorch library on a custom dataset. Once I got the output I was unsure what to make of it as the mAP score seems ...
0
votes
0
answers
42
views
KeyError: tensor(1) when trying to train Faster R-CNN model
I am getting the following error when trying to train my Faster R-CNN model. Can someone trace the source of the error or know what is wrong?:
The code for training:
!python train.py --data ...
- 11
0
votes
1
answer
27
views
Dimensions Output mismatch for Faster r cnn ResNet50 model
Am trying to make am classification and detection model with faster r cnn ResNet50, but I keep on getting this error
ValueError Traceback (most recent call last)
Cell In[...
- 23
1
vote
2
answers
241
views
Faster RCNN with Pytorch Lightning
Faster RCNN model has different behaviour for model.eval() and model.train().
At model.eval() mode model(images, targets) - returns prediction
At model.train() model model(images) return loss only
I'...
- 1,737
0
votes
0
answers
27
views
How to use label_smooth in Tensorflow object detection API?
How to use label_smooth in Tensorflow object detection API when I train Faster rcnn?
I try to put
label_smoothing = 0.1
factor = tf.constant(label_smoothing)
target_tensor *= (1 - factor)
...
- 3
0
votes
0
answers
68
views
Extermely slow Faster R-CNN training and memory usage increase every epoch
I am currently training Faster R-CNN algorithm with Mobilenetv2 as backbone. My own dataset is used, which approximately 9k images inlcuding train and valid set.
Initially, the training process goes ...
- 3
0
votes
0
answers
41
views
Offsets and label creation for object detection training/inference
I am new to object detection. I know that in object detection anchors are used for detecting objects in the feature map using offsets instead of predicting actual location of the object w.r.t input ...
1
vote
0
answers
173
views
How to define the forward pass in a custom PyTorch FasterRCNN?
I'm trying to write a custom PyTorch FasterRCNN that's based on an existing PyTorch model, fasterrcnn_resnet50_fpn, but I'm getting stuck on how to correctly write the forward pass. Following this ...
- 1,277
0
votes
0
answers
662
views
"error: command '/usr/local/cuda/bin/nvcc' failed with exit code 1" getting this error while building setup.py of bottom-up-attention.pytorch
so i am trying to extract features from images for my project using "https://github.com/MILVLG/bottom-up-attention.pytorch" and getting this error again and again.
I have successfully ...
- 1
0
votes
1
answer
69
views
While implementing mRCNN, experiencing issue with "No module named 'keras.engine'"
I am trying to implement mRCNN, and while trying to utilize it with Keras 2.0, I have repeatedly had the error:
Traceback (most recent call last):
File "C:\Users\Warre\PycharmProjects\test3\main....
0
votes
0
answers
66
views
Should we consider GeneralizedRCNNTransform while selecting anchor sizes in Faster R-CNN
I am training a Faster R-CNN on pytorch. The size of my images are 512x512 pixels. Until now, I computed the anchor sizes based on the sizes of my bounding box annotations on the 512x512 pixel images. ...
- 1
0
votes
0
answers
43
views
Bounding box is in misplaced
I'm currently engaged in refining Faster RCNN. I annotate my images using makesense.ai. Within these images, I focus solely on a single class. However, upon plotting the images and bounding boxes, I'...
- 1
0
votes
0
answers
116
views
How to use the pytorch-grad-cam library to make visualization for a specific tag for the Faster rcnn reset 50 fpn v2 model
I use EigenCAM from the library unfortunately, there is no example in the guide and documentation of how to make a visualization for a specific class. I tried to write my own function for the class ...
0
votes
0
answers
257
views
Training fasterrcnn_resnet50_fpn on visdrone dataset
I am trying to finetune the model fasterrcnn_resnet50_fpn from PyTorch on visdrone dataset for an Object Detection task. After some time from the beginning of the training loop, I got the following ...
0
votes
1
answer
24
views
Which algorithm to use to detect the appearance of sidebar on car multimedia booting?
I need to create a system for detecting the time that elapses between the start of booting the multimedia system in the car until the first appearance of the sidebar on the screen. Sidebar has a ...
0
votes
0
answers
77
views
Pytorch- Is it possible to apply self-attention into Faster R-CNN's ROI layer?
Everyone; I am trying to apply Self-attention mechanism in Faster R-CNN's ROI layers output. My task is to have Action Recognition; I have used different architectures from vanilla CNN + GRU to ...
- 39
0
votes
0
answers
17
views
Tensorflow Object Detection API - FasterCNN - Able to use normalised floats array stored in tfrecord
When using TensorFlow Object Detection API, and FasterCNN model from the Detection Zoo, are you able to store a normalised 2D array (0-1) as floats into a tfrecord and use this as model inputs?
I am ...
4
votes
2
answers
137
views
Pytroch segmentation model(.pt) not converting to CoreML
According to apple article link we need Wrap the Model to Allow Tracing which is be followed the same.
class WrappedDeeplabv3Resnet1011(nn.Module):
def __init__(self):
super(...
- 293
0
votes
1
answer
75
views
URLError: <urlopen error [WinError 10054] An existing connection was forcibly closed by the remote host>
I'm currently following along the PyTorch Tutorial here. As I call fasterrcnn_resnet50_fpn(), I'm getting the following error:
URLError: <urlopen error [WinError 10054] An existing connection was ...
- 71
0
votes
0
answers
27
views
Faster R-CNN Region Proposal Network loss sampling on FPN
I would like to know whether the sampling of positive/negative boxes during the loss calculation within the RPN of a Faster R-CNN w.r.t. to a FPN Backbone is done over all layers (i.e. if N=256, we ...
- 83
0
votes
0
answers
29
views
Faster RCNN pretrain model training error
Tensorflow : 2.10.1
Python 3.10.0
Pretrained model:faster_rcnn_inception_resnet_v2_640x640_coco17_tpu-8
When running the code
TRAINING_SCRIPT = os.path.join(paths['APIMODEL_PATH'], 'research', '...
0
votes
0
answers
20
views
Unique Detections from a Camera video feed using Pytorch Faster-RCNN Model
I have made a custom dataset then used that to fine tune Pytorch faster RCNN model. Now I have a conveyor, on top of which I have placed a camera. I am using the image from that camera and passing it ...
0
votes
1
answer
207
views
Tuned Faster R-CNN model detect maximum 100 objects, but there is 300+ objects on the image. How to increase this maximum?
I use this code to prepare model for optimization:
# load Faster RCNN pre-trained model
Faster_RCNN_tuned_model = torchvision.models.detection.fasterrcnn_resnet50_fpn(weights="DEFAULT")
# ...
1
vote
0
answers
116
views
Bounding box regression loss increases where RPN loss decreases Oriented RCNN
I am playing around with an oriented RCNN with a custom satellite DOTA dataset in MMRotate. When I train the Oriented RCNN, the bounding box regression loss starts very low (loss ~ 0.009) and almost ...
- 71
0
votes
0
answers
214
views
Faster RCNN class-wise output and the number of labels
I am applying pretrained Faster RCNN model on my custom dataset and wonder if there are any ways that I can look at the number of labels in all output images by class.
I am refering the github below:
...
- 1
0
votes
1
answer
575
views
No module named 'bbox'
I am working on one project which should be re written, so i did not write from beginning, and there is no opportunity to ask person who wrote the code.
I have next part of code:
from ..cython.bbox ...
- 15
1
vote
0
answers
110
views
Faster RCNN loss Nan from the beginning
I'm using Pytorch with Faster RCNN on dataset having 2 classes and about 100 images for training and 35 for validation in a multi node and multi gpu environment. Just to simplify the debugging, I'm ...
- 187
0
votes
1
answer
53
views
Faster R-CNN stabilize object regression
I've implememented a variant of Faster R-CNN and was wondering about the canonical way to stabilize the training of the (final) bounding box regression. At the moment, I have the issue, that the ...
- 83
0
votes
0
answers
71
views
Loss Function Sum or Mean Region Proposal Network Faster R-CNN
My Region Proposal Network loss function has a low loss.
Evaluate model: 2/2 [==============================] - 1s 30ms/step - loss: 4.3797e-04
However, the predictions it makes are very off. When I ...
0
votes
0
answers
74
views
Region Proposal Network Loss Function Faster R-CNN in Python
I have been trying to create a custom loss for the region proposal network for Faster R-CNN.
Here is my loss class:
class RPN_Loss(tf.keras.losses.Loss):
def L_cls(self,objectness_pred,...
1
vote
1
answer
954
views
Getting RuntimeError : configs/COCO-Detection/faster_rcnn_R_50_FPN_3x.yaml not available in Model Zoo! while Running Detectron2 for Object Detection
Want to train a custom Image Dataset by Detectron2 Faster_RCNN model . I am using a wsl2 ubuntu terminal and a VScode in my windows os. In my train.py, I initiate a config_file_path with "configs/...
0
votes
2
answers
580
views
2 Faster R-CNN models with different backbones, produce the same results during inference
Im using a python program from this tutorial. I've copied it as is and made changes to directories and label list so that the program takes in my dataset. The first Faster R-CNN model was trained with ...
- 11
0
votes
0
answers
28
views
How do I solve the problem of displaying accuracy parameter in CNN?
I'm facing a problem while computing accuracy parameter and sketching it in a graph.
I was expecting a drawn accuracy and loss graphs. In line:
model.compile(optimizer='adam', loss='...
0
votes
1
answer
743
views
ERROR: 'NoneType' object has no attribute 'shape'
image_data = []
for index in range(len(annotations_bbox[:10])):
img = cv2.imread('/content/drive/MyDrive/dataset/images/'+annotations_bbox['image_path'][index])
image_data.append(img)
height, ...
- 1
1
vote
0
answers
377
views
Is there a way to change the backbone of Detectron2 keypoint detection model with custom object detection model?
I am working on a task where I have to detect people and do activity recognition. I am using **Detectron2 **Keypoint detection model.There are many False negatives with the detection. So my solution ...
- 11
1
vote
0
answers
109
views
Error in checkpoint.py in Tensorflow2 on custom dataset using pre-built FRCNN model
I am using TF2 for teeth classification using panoramic X-ray images. The images have been reduced to 640x640 and the annotations scaled accordingly.
The original image looks like this:
The test run ...
- 162
0
votes
1
answer
769
views
How to implement a custom FastRCNNOutputLayers on Detectron2?
I need to implement a new architecture on the Faster-RCNN classifier on top of Faster-RCNN model implemented in Detectron2 framework. I need to implement both a new loss function on classification and ...
0
votes
1
answer
329
views
ROBOFLOW-tensorflow-object-detection-faster-rcnn_ cannot import name 'get_config' from 'tensorflow.python.eager.context'
I was trying to implement Fast RCNN model on my custom dataset using the google colab ROBOFLOW-tensorflow-object-detection-faster-rcnn.ipynb, but when I run the 'Install required packages' section ,...
- 1
1
vote
0
answers
95
views
Mask RCNN Keras custom image dimensions
I am trying to follow the example provided by https://tensorflow-object-detection-api-tutorial.readthedocs.io/en/latest/training.html#configure-the-training-pipeline to train the ...
- 11
0
votes
1
answer
550
views
Limited number of bounding boxes in fasterRCNN PyTorch
I use a pre-trained fasterRCNN from PyTorch to detect small obstacles in a 4k image. Currently I am facing some issues with false negatives, so objects are not detected.
I recognized that the number ...
0
votes
0
answers
261
views
PyTorch object detection tutorial problem (can't run training)
I'm working with Mask RCNN neural network and with PyTorch, I need to load dataset, and I followed this tutorial: https://pytorch.org/tutorials/intermediate/torchvision_tutorial.html I have ...
- 53
0
votes
1
answer
158
views
Pytorch model outputting labels it was not trained on
I have a Faster R CNN detector which I've trained with pytorch lightning on a quite noisy, but large, dataset. I would expect that after 1 epoch of training, the model would only output labels in the ...
- 97
0
votes
1
answer
68
views
`fasterRCNNLayers` for simple custom network give error message
When using fasterRCNNLayers for creating faster rcnn network from custom network, it gives error message for upgrading the network to faster rcnn network. matlab 2022a update 3 is used.
I have traced ...
- 4,067
1
vote
0
answers
265
views
Mask RCNN error - ValueError: shapes (5,1048576) and (3136,1) not aligned: 1048576 (dim 1) != 3136 (dim 0) after removing use_mini_mask
image_id = random.choice(dataset.image_ids)
print(image_id)
image, image_meta, gt_class_id, gt_bbox, gt_mask = modellib.load_image_gt(dataset,
config, image_id)
info = dataset.image_info[image_id]
...
- 11
1
vote
1
answer
4k
views
How do I create a model from a state dict?
I am trying to load a checkpoint pth file from the faster_rcnn_resnet101 model which is not currently in the PyTorch model zoo.
This causes PyTorch to throw a KeyError saying that I the layers in the ...
- 163
0
votes
1
answer
476
views
Detectron2 code not showing anything in object detection
I am using the code described in this article for running inference (object detection) on an image using a trained model.
# import some common detectron2 utilities
from detectron2.engine import ...
- 2,738
1
vote
0
answers
196
views
trouble using demo notebook available on hugging face visual bert with my own image
I'm having issues using my own image with this code. I would like to upload my own image rather than using a predefined URL link. For example, in this notebook, we are using the following code :
URL = ...
- 307
0
votes
1
answer
1k
views
How can I extract features from pytorch fasterrcnn_resnet50_fpn
I tried to extract features from following code.
However, it says 'FasterRCNN' object has no attribute 'features'
I want to extract features with (36, 2048) shape features when it has 36 classes.
Is ...
- 484