A Theory-Inspired Framework for Few-Shot Cross-Modal Sketch Person Re-Identification

Contribution

Uses alignment augmentation and knowledge transfer to bridge sketch/RGB re-ID with few labels.

Abstract

We propose a theory-inspired few-shot cross-modal sketch person re-identification framework that tackles the large sketch/RGB gap under scarce supervision. Alignment Augmentation and a Knowledge Transfer Catalyst are used to improve domain alignment and boost cross-modal generalization.

BibTeX

@inproceedings{gong2026ktcaa,
  title={A Theory-Inspired Framework for Few-Shot Cross-Modal Sketch Person Re-Identification},
  author={Gong, Yunpeng and Hou, Yongjie and Shi, Jiangming and Diep, Kim Long and Jiang, Min},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  year={2026},
  doi={10.1609/aaai.v40i6.42425},
}