Can u give some detail about training?

YudeWang / SEAM

Self-supervised Equivariant Attention Mechanism for Weakly Supervised Semantic Segmentation, CVPR 2020 (Oral)

MIT License

539 stars 97 forks source link

Can u give some detail about training? #29

Open liyiming001 opened 3 years ago

liyiming001 commented 3 years ago

Dear Wang, I just use the default training script to run train.py, However the result are lower than paper said(the result mIOU I test in VOC2012 val are 0.388), So can u give some detail about training?(some hyper-param?)

YudeWang commented 3 years ago

Hi @liyiming001, The result in paper is achieved by the default params setting. Please provide more details.

liyiming001 commented 3 years ago

I did nothing in train_SEAM.py except the voc_root, the othes are kept (such as the b_sz=8, lr=0.01, wt_dec=5e-4, weights=ilsvre...) , So I have no idea about this.

YudeWang commented 3 years ago

@liyiming001 What about score curve generated by evaluation.py?