Error in `resize` under multibackend postprocess when using custom detector about yolo_tracking HOT 8 CLOSED

Ok. So the error is here:

For the bboxes, make sure that:

• the format follows: [x1, y1, x2, y2, confidence, class]

For the crop:

• The crop is loaded correctly by displaying it using OpenCV's imshow function

from yolo_tracking.

Ok. So the error is here:

yolo_tracking/boxmot/appearance/reid_multibackend.py

Lines 201 to 205 in 80bc49b

	crop = cv2.resize(
	crop,
	(128, 256), # from (x, y) to (128, 256) | (w, h)
	interpolation=cv2.INTER_LINEAR,
	)

For the bboxes, make sure that:

• the format follows: [x1, y1, x2, y2, confidence, class]

For the crop:

• The crop is loaded correctly by displaying it using OpenCV's imshow function

Yes, i'm sure my input bboxes to format [x1, y1, x2, y2, confidence, class], because it's the same when I use it with ByteTrack and it doesn't give an error, about cv2 crop, I dont know how to fix, my cv2 can use imshow normally :((

from yolo_tracking.

Run imshow on the crop. After this line of code:

from yolo_tracking.

Run imshow on the crop. After this line of code:

yolo_tracking/boxmot/appearance/reid_multibackend.py

Line 199 in 80bc49b

crop = img[y1:y2, x1:x2]

I dont understand, What do you want to display with imshow function? pls :((

from yolo_tracking.

the crop. By visualizing the crop you will understand what the model is creating the embedding out of.

from yolo_tracking.

the crop. By visualizing the crop you will understand what the model is creating the embedding out of.

Oh, I understand. Thank you so much, my mistake in main.py :(((

from yolo_tracking.

I have another question, When I use YOLO to predict next location , it work and return the next location bounding box in predict func , but with img matching this usually not return the next location predict :(((

from yolo_tracking.

Please, create another issue if you want to change topic. Closing this down as it seems to be solved 😄

from yolo_tracking.