RED-Net

This repository is implementation of the "Image Restoration Using Very Deep Convolutional Encoder-Decoder Networks with Symmetric Skip Connections".
To reduce computational cost, it adopts stride 2 for the first convolution layer and the last transposed convolution layer.

Requirements

PyTorch
Tensorflow
tqdm
Numpy
Pillow

Tensorflow is required for quickly fetching image in training phase.

Results

Input	JPEG (Quality 10)

AR-CNN	RED-Net 10

RED-Net 20	RED-Net 30

Usages

Train

When training begins, the model weights will be saved every epoch.
If you want to train quickly, you should use --use_fast_loader option.

python main.py --arch "REDNet30" \  # REDNet10, REDNet20, REDNet30               
               --images_dir "" \
               --outputs_dir "" \
               --jpeg_quality 10 \
               --patch_size 50 \
               --batch_size 16 \
               --num_epochs 20 \
               --lr 1e-4 \
               --threads 8 \
               --seed 123 \
               --use_fast_loader

Test

Output results consist of image compressed with JPEG and image with artifacts reduced.

python example --arch "REDNet30" \  # REDNet10, REDNet20, REDNet30
               --weights_path "" \
               --image_path "" \
               --outputs_dir "" \
               --jpeg_quality 10