GFPGAN aims at developing a Practical Algorithm for Real-world Face Restoration.

Dependencies and Installation

• Python >= 3.7 (Recommend to use Anaconda or Miniconda)
• PyTorch >= 1.7
• Option: NVIDIA GPU + CUDA
• Option: Linux

Install basicsr We use BasicSR for both training and inference pip install basicsr An open-source image and video restoration toolbox https://github.com/XPixelGroup/BasicSR

Install facexlib We use face detection and face restoration helper in the facexlib package pip install facexlib A collection that provides useful face-relation functions https://github.com/xinntao/facexlib

If you want to enhance the background (non-face) regions with Real-ESRGAN, you also need to install the realesrgan package pip install realesrgan A practical algorithm for general image restoration https://github.com/xinntao/Real-ESRGAN

pip install -r requirements.txt python setup.py develop

Inference!

python quality_improvement.py -i data -o output -v 1.3 -s 2

Usage: python quality_improvement.py -i inputs/whole_imgs -o output -v 1.3 -s 2 [options]...

  -h                   show this help
  -i input             Input image or folder. Default: data/input_sample.jpeg
  -o output            Output folder. Default: output
  -v version           GFPGAN model version. Option: 1 | 1.2 | 1.3. Default: 1.3
  -s upscale           The final upsampling scale of the image. Default: 2
  -bg_upsampler        background upsampler. Default: realesrgan
  -bg_tile             Tile size for background sampler, 0 for no tile during testing. Default: 400
  -suffix              Suffix of the restored faces
  -only_center_face    Only restore the center face
  -aligned             Input are aligned faces
  -ext                 Image extension. Options: auto | jpg | png, auto means using the same extension as inputs. Default: auto