search

YudeWang / SEAM

Self-supervised Equivariant Attention Mechanism for Weakly Supervised Semantic Segmentation, CVPR 2020 (Oral)
MIT License
539 stars 97 forks source link

What is the pipeline to reproduce the result of Table 6 in your paper? #23

Closed Kyfafyd closed 3 years ago

Kyfafyd commented 3 years ago

Thanks for your excellent work and released code.

Looking forward to your reply!

YudeWang commented 3 years ago

Hi @Kyfafyd ,

  1. Train SEAM network to generate CAM
  2. Train AffinityNet basing on the results of step1 to generate pseudo labels
  3. Train Deeplab model by pseudo labels from step2.
Kyfafyd commented 3 years ago

Hi @Kyfafyd ,

  1. Train SEAM network to generate CAM
  2. Train AffinityNet basing on the results of step1 to generate pseudo labels
  3. Train Deeplab model by pseudo labels from step2.

Thanks very much. Besides, I wonder what is the pipeline to reproduce the result of Table 6 using your provided pretrained weights?

I just need to do step 3?

YudeWang commented 3 years ago

Hi @Kyfafyd ,

  1. Train SEAM network to generate CAM
  2. Train AffinityNet basing on the results of step1 to generate pseudo labels
  3. Train Deeplab model by pseudo labels from step2.

Thanks very much. Besides, I wonder what is the pipeline to reproduce the result of Table 6 using your provided pretrained weights?

I just need to do step 3?

This repository has provided model weights to generate pseudo labels and you only need to do step 3. And here is some suggestions for step 3: https://github.com/YudeWang/SEAM/issues/11

YudeWang commented 3 years ago

Code has release in https://github.com/YudeWang/semantic-segmentation-codebase/tree/main/experiment/seamv1-pseudovoc and I will close this issue.