search

hanyoseob / pytorch-noise2void

Noise2Void - Learning Denoising from Single Noisy Images
https://github.com/hanyoseob/pytorch-noise2void
109 stars 14 forks source link

Wrong Test Results #2

Open GuoLanqing opened 4 years ago

GuoLanqing commented 4 years ago

I trained the model according to your methods mentioned in readme. Then, I want to test the result of denoising. But the generated denoised image is almost gray. Is there something wrong? And why the sample ratio in test process is 1?

ZhidongYang commented 4 years ago

I had the same question. All of the test results are almost gray images. Is there something wrong?

bensterl15 commented 3 years ago

I had the same problem (with grayscale images). Setting the number of channels to 1, normalizing my raw data, and commenting out the colormaps in train.py (starting at line 415) solved it for me.

weihua9217 commented 2 years ago

I had the same problem. In my case, I didn't load the pretrained model correctly. Replacing "--dir_checkpoint checkpoint", by "--dir_checkpoint checkpoints" may solve it.

xiaowowwwww commented 1 year ago

Hello, I would like to know the reason why after modifying the parameters set in the readme, I have been unable to test my image. Thank you very much!