l2d_sam_iter2

See-through: Body-Part Parsing (SAM)

The body-part parsing model of See-through: SAM-HQ (ViT-H) fine-tuned for semantic segmentation of anime characters into 19 body parts. The weights are in checkpoint-18000.pt.

See inference/demo/bodypartseg_sam.ipynb in our GitHub repository for an interactive demo, and inference/scripts/parse_live2d.py for how the model is loaded.

Licence

These weights are licensed under Apache-2.0, as is the See-through code. They are fine-tuned from SAM-HQ, which is based on Meta's Segment Anything; both are also Apache-2.0. Commercial use is permitted. See LICENSE for the licence text and NOTICE for attributions and the changes we made.

Citation

If you find this work useful, please cite:

@inproceedings{lin2026seethrough,
  author={Lin, Jian and Li, Chengze and Qin, Haoyun and Chan, Kwun Wang and Jin, Yanghua and Liu, Hanyuan and Choy, Stephen Chun Wang and Liu, Xueting},
  title={See-through: Single-image Layer Decomposition for Anime Characters},
  booktitle={Proceedings of the Special Interest Group on Computer Graphics and Interactive Techniques Conference Conference Papers},
  series={SIGGRAPH Conference Papers '26},
  publisher={Association for Computing Machinery},
  address={New York, NY, USA},
  year={2026},
  pages={1--11},
  doi={10.1145/3799902.3811209},
  url={https://doi.org/10.1145/3799902.3811209}
}