Attention-Guided Hierarchical Structure Aggregation for Image Matting

This is the official implementation of HAttMatting(CVPR2020)

Code and paper is coming soon!!!

News

0. The dataset link is : Distinctions-646.
   
1. 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.
2. The thumbnail from our dataset have been uploaded.

   Training Image	Training GT

3. Our project Page is now fully open to the public, including pipelines, data set examples, etc. Paper will be open in two days.
   
4. The file for generating the composition dataset has been uploaded. It can synthesize data three times faster than the original code, which can save about two days of waiting!!!

Project Page

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

Visual Results

Input Image	Our Result

Dataset

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.

Citation

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}
}