search

CVI-SZU / CCAM

[CVPR 2022] C2AM: Contrastive learning of Class-agnostic Activation Map for Weakly Supervised Object Localization and Semantic Segmentation
MIT License
190 stars 27 forks source link

The size of x1, x2 not match #20

Closed jokingww closed 1 year ago

jokingww commented 1 year ago

I just read your paper and code, it's a great work. I'm confused in code 'model.py',

    x1 = self.layer3(x)
    x2 = self.layer4(x1)
    return torch.cat([x2, x1], dim=1)

after downsample, the size of x2, x1 are not match. Did I understand something wrong?

Sierkinhane commented 1 year ago

Hi, we have changed the the last stride of resnet block. More details can be found at here.

jokingww commented 1 year ago

I didn't notice this change. Thanks for explaination.