gqf126 / AIRNet

Code for TIP 2021 paper "Real-Time Deep Image Retouching Based on Learnt Semantics Dependent Global Transforms"

AIRNet

Code for TIP 2021 paper: Real-Time Deep Image Retouching Based on Learnt Semantics Dependent Global Transforms

Dependencies and Installation

• Python 3 (Recommend to use Anaconda)
• PyTorch >= 1.0
• NVIDIA GPU + CUDA
• Python packages: pip install numpy opencv-python lmdb pyyaml
• TensorBoard:
  • PyTorch >= 1.1: pip install tb-nightly future
  • PyTorch == 1.0: pip install tensorboardX

Datasets

MIT-Adobe FiveK dataset

We also recommand to use the preprocessed datasets MIT-Adobe FiveK preprocessed dataset (kindly shared by Jingwen He, author of CSRNet)

Inference

1. Modify the configuration file options/test/test_Meta_Y_UV_Para.json.
2. Run command:

python test.py -opt options/test/test_Meta_Y_UV_Para.json

Acknowledgement

• This code is based on mmsr.

BibTex

@article{gao2021real,
  title={Real-time deep image retouching based on learnt semantics dependent global transforms},
  author={Gao, Qifan and Wu, Xiaolin},
  journal={IEEE Transactions on Image Processing},
  volume={30},
  pages={7378--7390},
  year={2021},
  publisher={IEEE}
}