Practical Wide-Angle Portraits Correction with Deep Structured Models

December 19, 2022 · View on GitHub

Jing Tan, Shan Zhao, Pengfei Xiong, Jiangyu Liu, Haoqiang Fan, Shuaicheng Liu

[arxiv] [BibTeX]

Note

In this repository, we will release dataset and pytorch implementation of our paper.

Quick Start

Python 3.6+, Pytorch 1.2, torchvision 0.4, cuda10.0, and other requirements.

All codes are tested on Linux.

Installation

Clone the repository

git clone https://github.com/TanJing94/Deep_Portraits_Correction.git
cd Deep_Portraits_Correction

Install dependencies
```
pip install -r requirements.txt
```

Resources preparation

Dataset

Download from [BaiduCloud], extraction code: 5pe5
Please refer to [dataset.py] for data usage

Pre-trained model

Training

Testing

Citation

If you find this work or code is helpful in your research, please cite:

@inproceedings{tan2021practical,
  title={Practical Wide-Angle Portraits Correction with Deep Structured Models},
  author={Tan, Jing and Zhao, Shan and Xiong, Pengfei and Liu, Jiangyu and Fan, Haoqiang and Liu, Shuaicheng},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={3498--3506},
  year={2021}
}