meteorshowers / RCF-pytorch

Richer Convolutional Features for Edge Detection model in pytorch CVPR2017
509 stars 145 forks source link

Why my debug image output is all black? #27

Open Gengzigang opened 5 years ago

Gengzigang commented 5 years ago

sss Is it normal? The last three output maps are always all black after many epoches.

LIMENG0307 commented 4 years ago

I met the same situation,did you know it。

Gengzigang commented 4 years ago

I met the same situation,did you know it。

Maybe the learning rate is too high.

khflam commented 4 years ago

I met the same situation,did you know it。

Maybe the learning rate is too high.

Hi Gengzigang, did you manage to fix the black images issue? Just curious whether the learning rate was the cause.

LunarSun commented 4 years ago

sss Is it normal? The last three output maps are always all black after many epoches.

I met the same problem,Have you solved it?

zcong17huang commented 3 years ago

sss Is it normal? The last three output maps are always all black after many epoches.

I met the same problem as you, Have you solved it? Please tell me.

zcong17huang commented 3 years ago

I think i have found the solution. Change the kernel size of upsample layer5 from 32 to 16, the problem will be solved.