search

vasgaowei / TS-CAM

Codes for TS-CAM: Token Semantic Coupled Attention Map for Weakly Supervised Object Localization.
https://openaccess.thecvf.com/content/ICCV2021/papers/Gao_TS-CAM_Token_Semantic_Coupled_Attention_Map_for_Weakly_Supervised_Object_ICCV_2021_paper.pdf
Apache License 2.0
137 stars 25 forks source link

Use the pretrained model, can't achieve the results of the paper #8

Open lingxinglovecode opened 2 years ago

lingxinglovecode commented 2 years ago

Use the pretrained TS-CAM model to evaluate on ILSVRC, but get the following results

image

vasgaowei commented 2 years ago

Hi, thanks for your attention. The localization accuracies are a little lower than those in the paper. Maybe you can try to train the model yourself, seeing the final results.

abhayraw1 commented 2 years ago

I can also confirm this. I ran the evaluation script with the checkpoint provided and was getting similar results!

image