jbeomlee93 / AdvCAM

Anti-Adversarially Manipulated Attributions for Weakly and Semi-Supervised Semantic Segmentation (CVPR 2021)
MIT License
126 stars 16 forks source link

FileNotFoundError: [Errno 2] No such file or directory: 'result/cam_adv_mask/2007_000032.npy' #5

Closed seyeeet closed 3 years ago

seyeeet commented 3 years ago

Thanks for sharing your code. after downloading all the materials now I am running the bash get_mask_quality.sh but it is super slow and I am wondering how long that process should take? and what does that _work function in obtain_CAM_masking.py really do? do I need to do it?

jbeomlee93 commented 3 years ago

I think it depends on your machine, but it can be a bit slow because adversarial climbing is iteratively done by 27 times for 10,582 images. It took around 14 hours for me with 4 GPUs.

If you want to evaluate the quality of the initial seed shown in Table 1, you can just use get_seed_quality.sh, which operates only on 1,456 training images. However, if you want to get the final pseudo mask, you should run get_mask_quality.sh for the full 10,582 images.

The _work function is for multi-GPU processing.

Thanks.

seyeeet commented 3 years ago

Thanks