search

clovaai / wsolevaluation

Evaluating Weakly Supervised Object Localization Methods Right (CVPR 2020)
MIT License
330 stars 55 forks source link

Sequence of normalization and resize of CAM? #54

Open Kevinz-code opened 2 years ago

Kevinz-code commented 2 years ago

Hi, Thanks for your good work.

I'm a little wondering the line86-line88 in your inference.py. '' cam_resized = cv2.resize(cam, image_size, interpolation=cv2.INTER_CUBIC) cam_normalized = normalize_scoremap(cam_resized) ''

In WSOL, after we get a certain class's CAM score of the feature size (hxw, e.g, 7x7), do we resize it to the original image (224x224) and then normalize the score to [0, 1], or do we normalize the CAM score (in the 7x7 shape) to [0, 1] and then resize it to the original image?

I'm looking forward to your reply.

Thanks in advance!

Best,

coallaoh commented 2 years ago

Thanks for the question!

There was no standard practice for the ordering between resize and normalization before. We have decided to fix them to resize --> normalize in our inference code. We do not think the ordering to critically change the evaluation results though.