Yuhao Liu*, Jiake Xie*, Xiao Shi, Yu Qiao, Yujie Huang, Yong Tang, Xin Yang
This is the official PyTorch implementation of our paper Tripartite Information Mining and Integration for Image Matting that has been accepted to 2021 IEEE/CVF International Conference on Computer Vision (ICCV 2021).
The repository is trained and tested on Ubuntu 18.04.3 LTS, based on the main following settings.
- python 3.6+
- Pytorch 1.1.0+
- cuda 11.0+
To quickly test sample images with our model (trained on Adobe dataset), you can just run through
cd Timi-Net
python test.py By default, the code takes the data in the "./inputs/" folder, loads the "TIMI-Net.pth" model and saves results in the "./outputs/" folder. Please read the code to see other parameter settings.
| Datasets | Results | Model |
|---|---|---|
| Adobe Composition-1K | Google Drive / Baidu Drive(code: jhmr) | Google Drive |
| Distinctions-646 | is coming | is coming |
| Human-2K | is coming | is coming |
| Datasets | Results |
|---|---|
| Microsoft One Drive | Baidu Drive(code: fvim) |
This project is only for research purpose. For any other questions, please feel free to contact us.
This repository highly depends on the GCA-matting repository at https://github.com/Yaoyi-Li/GCA-Matting. We thank the authors of GCA for their great work and clean code.
If you use this code for your research, please cite our paper.
@InProceedings{Liu_2021_ICCV,
author = {Liu, Yuhao and Xie, Jiake and Shi, Xiao and Qiao, Yu and Huang, Yujie and Tang, Yong and Yang, Xin},
title = {Tripartite Information Mining and Integration for Image Matting},
booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
month = {October},
year = {2021},
pages = {7555-7564}
}