zonghaofan / noise2noise

Remove watermark.By adding dilation rate, get better results.

1.noise2noiset-pytorch

my chinese blog:https://blog.csdn.net/fanzonghao/article/details/109389708

2.train

python train_fzh.py

3.torch inference

img:python test_fzh.py

video:python test_video_model_fzh.py

4.requirements

pytorch1.5

torchvision0.6

cuda9.0+

5.some examples