图像去噪Noise2Void（N2V）的Pytorch复现代码，基于U-Net模型实现，原理详解，注释详细，包含训练好的模型

# RED-Net This repository is implementation of the "Image Restoration Using Very Deep Convolutional Encoder-Decoder Networks with Symmetric Skip Connections". <br /> To reduce computational cost, it adopts stride 2 for the first convolution layer and the last transposed convolution layer. ## Requirements - PyTorch - tqdm - Numpy - Pillow ## Results sigma=10，REDNet10： ![07_plt_x[10]_REDNet10](./data/07_plt_x[10]_REDNet10.png) ![08_plt_x[10]_REDNet10](./data/08_plt_x[10]_REDNet10.png) ## Usages ### Train ```bash python main.py --arch "REDNet30" \ # REDNet10, REDNet20, REDNet30 --images_dir "" \ --outputs_dir "" \ --gaussian_noise_level 10 \ --patch_size 50 \ --batch_size 16 \ --num_epochs 1000 \ --lr 1e-4 \ --threads 8 \ --seed 123 \ ``` ### Test ```bash python example --arch "REDNet30" \ # REDNet10, REDNet20, REDNet30 --weights_path "" \ --image_path "" \ --outputs_dir "" \ --gaussian_noise_level 10 ```