GSoC'22 Multi-tasking computer vision model: object detection, object segmentation and human pose detection #76
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
fengyuentau
changed the title
|
Please read comments carefully below:
|
Closed
fengyuentau
reviewed
Aug 8, 2022
fengyuentau
reviewed
Sep 3, 2022
models/multitask_centernet/demo.py
Outdated
Comment on lines
14
to
30
| if __name__ == "__main__": | ||
| parser = argparse.ArgumentParser() | ||
| parser.add_argument('--imgpath', type=str, default='images/d2645891.jpg', help="image path") | ||
| parser.add_argument('--modelpath', type=str, default='MCN.onnx') | ||
| args = parser.parse_args() | ||
|
|
||
| mcn = MCN(args.modelpath) | ||
| srcimg = cv2.imread(args.imgpath) | ||
| srcimg = mcn.detect(srcimg) | ||
| cv2.imwrite('result.png', srcimg) | ||
|
|
||
|
|
||
| # winName = 'using MCN in OpenCV' | ||
| # cv2.namedWindow(winName, 0) | ||
| # cv2.imshow(winName, srcimg) | ||
| # cv2.waitKey(0) | ||
| # cv2.destroyAllWindows() |
There was a problem hiding this comment.
A demo with webcam stream as input should be provided as well. You can find a example here.
Comment on lines
+176
to
+179
| img, newh, neww, padh, padw = self.resize_image(srcimg) | ||
| blob = cv2.dnn.blobFromImage(img, scalefactor=1 / 255.0, swapRB=True) | ||
| # blob = cv2.dnn.blobFromImage(self.preprocess(img)) | ||
| # Sets the input to the network |
There was a problem hiding this comment.
Move these lines into preprocess and call preprocess.
Comment on lines
+185
to
+202
| # inference output | ||
| row_ind = 0 | ||
| for i in range(self.nl): | ||
| h, w = int(self.inpHeight / self.stride[i]), int(self.inpWidth / self.stride[i]) | ||
| length = int(self.na * h * w) | ||
| if self.grid[i].shape[2:4] != (h, w): | ||
| self.grid[i] = self._make_grid(w, h) | ||
|
|
||
| outs[row_ind:row_ind + length, 0:2] = (outs[row_ind:row_ind + length, 0:2] * 2. - 0.5 + np.tile( | ||
| self.grid[i], (self.na, 1))) * int(self.stride[i]) | ||
| outs[row_ind:row_ind + length, 2:4] = (outs[row_ind:row_ind + length, 2:4] * 2) ** 2 * np.repeat( | ||
| self.anchor_grid[i], h * w, axis=0) | ||
|
|
||
| self.num_coords = outs.shape[1] - self.last_ind | ||
| outs[row_ind:row_ind + length, self.last_ind:] = outs[row_ind:row_ind + length, self.last_ind:] * 4. - 2. | ||
| outs[row_ind:row_ind + length, self.last_ind:] *= np.tile(np.repeat(self.anchor_grid[i], h * w, axis=0), (1, self.num_coords//2)) | ||
| outs[row_ind:row_ind + length, self.last_ind:] += np.tile(np.tile(self.grid[i], (self.na, 1)) * int(self.stride[i]), (1, self.num_coords//2)) | ||
| row_ind += length |
There was a problem hiding this comment.
Move these lines into postprocess
| cv2.putText(frame, label, (left, top - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), thickness=1) | ||
| return frame | ||
|
|
||
| def detect(self, srcimg): |
There was a problem hiding this comment.
Rename this function to infer and keep it simple like the following:
def infer(self, image):
input_blob = self.preprocess(image)
self.model.setInput(input_blob)
output_blob = self.model.forward(self.model.getUnconnectedOutLayersNames())
results = self.postprocess(output_blob)
return resultsHere is another example for reference.
fengyuentau
reviewed
Sep 5, 2022
There was a problem hiding this comment.
Please make demo work as soon as possible. I got the file missing error:
Traceback (most recent call last):
File "/Some/Path/opencv_zoo/models/multitask_centernet/demo.py", line 20, in <module>
mcn = MCN(args.modelpath)
File "/Some/Path/opencv_zoo/models/multitask_centernet/multitask_centernet.py", line 13, in __init__
with open('crowd_class.names', 'rt') as f:
FileNotFoundError: [Errno 2] No such file or directory: 'crowd_class.names'
fengyuentau
reviewed
Sep 7, 2022
Comment on lines
+109
to
+112
| person_indices = cv2.dnn.NMSBoxes(person_boxes, person_confidences, config['person_conf_thres'], | ||
| config['person_iou_thres']).flatten() | ||
| kp_indices = cv2.dnn.NMSBoxes(kp_boxes, kp_confidences, config['kp_conf_thres'], | ||
| config['kp_iou_thres']).flatten() |
There was a problem hiding this comment.
NMSBoxes returns a empty tuple if no person is detected. Calling flatten on an empty tuple triggers error.
Traceback (most recent call last):
File "/path/opencv_zoo/models/multitask_centernet/demo.py", line 14, in <module>
srcimg = mcn.detect(srcimg)
File "/path/opencv_zoo/models/multitask_centernet/multitask_centernet.py", line 207, in detect
srcimg = self.postprocess(srcimg, outs, padsize=(newh, neww, padh, padw))
File "/path/opencv_zoo/models/multitask_centernet/multitask_centernet.py", line 112, in postprocess
kp_indices = cv2.dnn.NMSBoxes(kp_boxes, kp_confidences, config['kp_conf_thres'],
AttributeError: 'tuple' object has no attribute 'flatten'
Please doublecheck with images with no person and even no objects at all.
|
You need to properly handle the case when there is no person in the image. Currently your script does not produce any boxes if there is no person in the image, such as the image below.
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.