WESPE: Weakly Supervised Photo Enhancer for Digital Cameras

TensorFlow implementation of "WESPE: Weakly Supervised Photo Enhancer for Digital Cameras"

First step

• Download the pre-trained VGG-19 model and put it into vgg_pretrained/ folder
• Download DPED dataset (patches for CNN training) and extract it into dped/ folder.

Train the model

• python train_model.py model= or sh train.sh

Training results

test PSNR for iphone is about 18 dB, similar to the original paper, 18.11 dB