This is the official implementation of HAttMatting(CVPR2020)

Code and paper is coming soon!!!

0.The data set is now fully open to the public. If you use it for non-commercial uses, please send us an email to access the link if necessary.

1. The thumbnail from our dataset have been uploaded.

Training Image	Training GT

Our Project Page is available now: HAttMatting. The more contents is updating.

Input Image	Our Result

Our Dinstinctions-646 are composed of 646 foreground images with manually annotated alpha mattes. We will release this dataset for encouraging future research on trimap-free image matting. We will not post a public download link and if you need this dataset for academic research, please send us a email for getting it.

If you use this code and dataset for your research, please cite our paper:

@InProceedings{Qiao_2020_CVPR,
    author = {Qiao, Yu and Liu, Yuhao and Yang, Xin and Zhou, Dongsheng and Xu, Mingliang and Zhang, Qiang and Wei, Xiaopeng},
    title = {Attention-Guided Hierarchical Structure Aggregation for Image Matting},
    booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
    month = {June},
    year = {2020}
}