search

CVI-SZU / CLIMS

[CVPR 2022] CLIMS: Cross Language Image Matching for Weakly Supervised Semantic Segmentation
MIT License
120 stars 12 forks source link

Need Coco baseline scores #18

Closed rB080 closed 1 year ago

rB080 commented 1 year ago

Hi!

We were trying to run the Coco codes, have a mIOU ~0.34. I am not sure if this is what we should expect, so it would be great if you could add baseline results to the repository. We did not change the default tuning settings.

Thanks

Sierkinhane commented 1 year ago

~35 mIoU for this repository. More and better results will be updated in the next version of CLIMS.

rB080 commented 1 year ago

I think that's consistent with what we are getting. Thanks a lot!

Bala93 commented 1 year ago

Hi @ProjectDisR Are you saying that you got miou of ~0.41 ? If yes, is this for pseudolabels of train or valid ? LIke you said, I also assume to use the created pseudolabels to train the deeplab and get the validation miou scores ? Let me know what you think.