xmed-lab/URN

About generate mask

ZechengLi19 opened this issue · 1 comments

Thanks for your perfect work! Now I have a question. If I want to generate WSSS mask for coco dataset? Can I use the weight you provide directly? Which weight and config file, I should use? for res2net.
Thanks for helping!!!

The intermediate weights are trained from voc dataset, and the weights of coco are missed.
You could train the segmentation model via the provided pseudo-mask from classification phase.