search

xtudbxk / DSRG-tensorflow

a tensorflow version for DSRG (Weakly-Supervised Semantic Segmentation Network with Deep Seeded Region Growing)
75 stars 20 forks source link

About Loss #27

Closed harrywang725 closed 4 years ago

harrywang725 commented 4 years ago

I tried to implement in pytorch , but during training , i met a problem that the BalancedSeedLoss will make softmax probability of one in all position , and it does not work in fg area ,and the loss will keep 8~9 . Have you met this problem? how could I solve this problem? thx.

xtudbxk commented 4 years ago

I didn't met the problem. And maybe you can compute the balancedseedloss for a specified image and display all the output of middle steps to see what happened.

harrywang725 commented 4 years ago

Thx, I used a single image to tried , I found something wrong with my weight initialize . When I fixed it , it's all right .

eyoujun commented 4 years ago

I meet the same question, would you please tell me how you fix it