🌺ICCV 2025🌺 CompleteMe: Reference-based Human Image Completion

Yu-Ju Tsai, Brian Price, Qing Liu, Luis Figueroa, Daniil Pakhomov, Zhihong Ding, Scott Cohen, Ming-Hsuan Yang
University of California, Merced - Adobe Research

Given occluded human images, CompleteMe can preserve identical and fine-detail information from the reference image and generate a consistent result.

News ✨ ✨

• [2025-06-25] Our paper is accepted by ICCV 2025.

Citation

If you find our work useful for your project, please consider citing our paper.

@article{tsai2025completeme,
        title={CompleteMe: Reference-based Human Image Completion},
        author={Tsai, Yu-Ju and Price, Brian and Liu, Qing and Figueroa, Luis and Pakhomov, Daniil and Ding, Zhihong and Cohen, Scott and Yang, Ming-Hsuan},
        journal={arXiv preprint arXiv:2504.20042},
        year={2025}
      }