Elamur Research⌕
☰
home / papers / 2026 / Iterative Self-Correction for Text-Driven Person Re-Identification with Large Vision-Language Models

Iterative Self-Correction for Text-Driven Person Re-Identification with Large Vision-Language Models

Paper · 2026 · Guijin Luo, Zequn Xie, Sihang Cai et al.

Summary

The paper addresses person re-identification (ReID), where systems that use vision-language models usually turn a query image into a description once and then retrieve with it. Two weaknesses follow. Converting an image to text drops identity details, and early hallucinated attributes cannot be undone. The authors propose Auto-ReID, which treats retrieval as a closed loop of reasoning and revision. A general VLM is first specialised with Hierarchical Progressive Tuning, a two-stage LoRA curriculum built from the Market-1501 and MSMT17 training splits:

  • Stage 1 trains structured attribute descriptions.
  • Stage 2 trains pairwise verification, attribute Q&A and corrective-feedback generation.

At inference, three modules work in a loop:

  • a Reasoner writes or refines a structured query;
  • a Hybrid Retriever mixes the changing text embedding with a fixed visual embedding of the query;
  • a Corrector checks the top candidates and sends back negative constraints.

On MSMT17 the full system reaches 89.2% mAP. The authors report the largest gains under occlusion. The work matters because it lets a ReID system correct itself without a human in the loop.

Key points

  • On MSMT17, the full Auto-ReID reaches 89.2% mAP and 91.8% Rank-1.
  • In the MSMT17 ablation:
    • a text-only loop drops to 66.5% mAP, which the authors attribute to semantic drift;
    • a static hybrid without the loop reaches 76.8%;
    • removing the tuning gives 74.2%.
    • The paper's hyperparameter table lists the no-iteration baseline at 85.8%, so the reported figures are inconsistent.
  • Hybrid similarity weighs visual against CLIP text features with α = 0.65, favouring the visual side. The Corrector inspects K = 20 candidates.
  • Gains level off at three refinement iterations, with the first iteration giving the biggest jump. Compute cost grows linearly with the number of iterations.
  • Code is available at github.com/GridNexus/Auto-ReID. The authors list as limitations the latency for real-time use and dependence on a clean visual anchor.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓National University of SingaporeSingapore · Singapore180 works · 34,304 citations🎓Zhejiang UniversityHangzhou · China172 works · 12,988 citations

 Fields, topics and concepts

📚 Shelves 2The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksComputer Vision and Pattern Recognition · 2,869 works

🧭 Topics 2Research subjects and languages the work is about.Language models & pretraining · 3,320 worksVision & language · 657 works

Cite

Cite as Guijin Luo, Zequn Xie, Sihang Cai, Chuxin Wang, Zhou Zhao and Yixuan Tang. 2026. Iterative Self-Correction for Text-Driven Person Re-Identification with Large Vision-Language Models. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.312

@inproceedings{luo2026iterative,
  title = {{Iterative Self-Correction for Text-Driven Person Re-Identification with Large Vision-Language Models}},
  author = {Guijin Luo, Zequn Xie, Sihang Cai, Chuxin Wang, Zhou Zhao and Yixuan Tang},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.312},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.312}
}

References

42 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Vaibhav Bansal, Gian Luca Foresti, and Niki Martinel. 2022. Cloth-changing person re-identification with self-attention. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 602–610. find
  2. Ruibing Hou, Bingpeng Ma, Hong Chang, Xinqian Gu, Shiguang Shan, and Xilin Chen. 2019. Interaction-and-aggregation network for person re-identification. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9317– 9326. find
  3. Marco Filax and Frank Ortmeier. 2021. On the influence of viewpoint change for metric learning. In 2021 17th International Conference on Machine Vision and Applications (MVA), pages 1–4. IEEE. find
  4. Xin Jin, Cuiling Lan, Wenjun Zeng, Guoqiang Wei, and Zhibo Chen. 2020. Semantics-aligned representation learning for person re-identification. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 11173–11180. find
  5. Lihua Fu, Yubin Du, Yu Ding, Dan Wang, Hanxu Jiang, and Haitao Zhang. 2022. Domain adaptive learning with multi-granularity features for unsupervised person re-identification. Chinese Journal of Electronics, 31(1):116–128. find
  6. Siyuan Li, Li Sun, and Qingli Li. 2023. Clip-reid: exploiting vision-language model for image re-identification without concrete text labels. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, pages 1405–1413. find
  7. Teng Fu, Haiyang Yu, Ke Niu, Bin Li, and Xiangyang Xue. 2025. Foundation model driven appearance extraction for robust multiple object tracking. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 3031–3039. find
  8. Wei Li, Rui Zhao, Tong Xiao, and Xiaogang Wang. 2014. Deepreid: Deep filter pairing neural network for person re-identification. In CVPR. find
  9. Xinqian Gu, Hong Chang, Bingpeng Ma, Shutao Bai, Shiguang Shan, and Xilin Chen. 2022. Clothes-changing person re-identification with rgb modality only. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 1060–1069. find
  10. Wei Li, Xiatian Zhu, and Shaogang Gong. 2018. Harmonious attention network for person re-identification. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2285–2294. find
  11. Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2024. Visual instruction tuning. Advances in neural information processing systems, 36.
  12. Peini Guo, Hong Liu, Jianbing Wu, Guoquan Wang, and Tao Wang. 2023. Semantic-aware consistency network for cloth-changing person re-identification. In Proceedings of the 31st ACM International Conference on Multimedia, pages 8730–8739. find
  13. Jiaxu Miao, Yu Wu, Ping Liu, Yuhang Ding, and Yi Yang. 2019. Pose-guided feature alignment for occluded person re-identification. In Proceedings of the IEEE International Conference on Computer Vision, pages 542–551. find
  14. Shuting He, Hao Luo, Pichao Wang, Fan Wang, Hao Li, and Wei Jiang. 2021. Transreid: Transformer-based object re-identification. In Proceedings of the IEEE/CVF international conference on computer vision, pages 15013–15022. find
  15. Xin Ning, Ke Gong, Weijun Li, Liping Zhang, Xiao Bai, and Shengwei Tian. 2020. Feature refinement and filter network for person re-identification. IEEE Transactions on Circuits and Systems for Video Technology, 31(9):3391–3402. find
  16. Weizhen He, Yiheng Deng, Shixiang Tang, Qihao Chen, Qingsong Xie, Yizhou Wang, Lei Bai, Feng Zhu, Rui Zhao, Wanli Ouyang, and 1 others. 2024. Instruct-reid: A multi-purpose person re-identification task with instructions. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog… find
  17. Ke Niu, Haiyang Yu, Xuelin Qian, Teng Fu, Bin Li, and Xiangyang Xue. 2025a. Synthesizing efficient data with diffusion models for person re-identification pre-training. Machine Learning, 114(3):1–25. find
  18. Weizhen He, Yiheng Deng, Yunfeng Yan, Feng Zhu, Yizhou Wang, Lei Bai, Qingsong Xie, Rui Zhao, Donglian Qi, Wanli Ouyang, and 1 others. 2025. Instruct-reid++: Towards universal purpose instruction-guided person re-identification. IEEE Transactions on Pattern Analysis and Machine Intelligence. find
  19. Peixian Hong, Tao Wu, Ancong Wu, Xintong Han, and Wei-Shi Zheng. 2021. Fine-grained shape-appearance mutual learning for cloth-changing person re-identification. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10513–10522. find
  20. Ke Niu, Haiyang Yu, Mengyang Zhao, Teng Fu, Siyang Yi, Wei Lu, Bin Li, Xuelin Qian, and Xiangyang Xue. 2025b. Chatreid: Open-ended interactive person retrieval via hierarchical progressive tuning for vision language models. ArXiv, abs/2502.19958. — [arXiv:2502.19958](https://arxiv.org/abs/2502.19958) find
  21. Xuelin Qian, Wenxuan Wang, Li Zhang, Fangrui Zhu, Yanwei Fu, Tao Xiang, Yu-Gang Jiang, and Xiangyang Xue. 2020a. Long-term cloth-changing person re-identification. In Proceedings of the Asian Conference on Computer Vision. find
  22. Xuelin Qian, Wenxuan Wang, Li Zhang, Fangrui Zhu, Yanwei Fu, Tao Xiang, Yu-Gang Jiang, and Xiangyang Xue. 2020b. Long-term cloth-changing person re-identification. arXiv preprint arXiv:2005.12633. — [arXiv:2005.12633](https://arxiv.org/abs/2005.12633) find
  23. Yifan Sun, Liang Zheng, Yi Yang, Qi Tian, and Shengjin Wang. 2018. Beyond part models: Person retrieval with refined part pooling (and a strong convolutional baseline). In ECCV, pages 480–496. find
  24. Hongchen Tan, Xiuping Liu, Yuhao Bian, Huasheng Wang, and Baocai Yin. 2021. Incomplete descriptor mining with elastic loss for person re-identification. IEEE Transactions on Circuits and Systems for Video Technology, 32(1):160–171. find
  25. Shixiang Tang, Cheng Chen, Qingsong Xie, Meilin Chen, Yizhou Wang, Yuanzheng Ci, Lei Bai, Feng Zhu, Haiyang Yang, Li Yi, and 1 others. 2023. Humanbench: Towards general human-centric perception with projector assisted pretraining. In Proceedings of the IEEE/CVF Conference on Computer Vision and Patt… find
  26. Michael Tschannen, Alexey Gritsenko, Xiao Wang, Muhammad Ferjad Naeem, Ibrahim Alabdulmohsin, Nikhil Parthasarathy, Talfan Evans, Lucas Beyer, Ye Xia, Basil Mustafa, Olivier Hénaff, Jeremiah Harmsen, Andreas Steiner, and Xiaohua Zhai. 2025. Siglip 2: Multilingual vision-language encoders with improv… find
  27. Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, and 1 others. 2024a. Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution. arXiv preprint arXiv:2409.12191.
  28. Qizao Wang, Xuelin Qian, Bin Li, Lifeng Chen, Yanwei Fu, and Xiangyang Xue. 2024b. Content and salient semantics collaboration for cloth-changing person re-identification. arXiv preprint arXiv:2405.16597. — [arXiv:2405.16597](https://arxiv.org/abs/2405.16597) find
  29. Weiyun Wang, Zhangwei Gao, Lixin Gu, Hengjun Pu, Long Cui, Xingguang Wei, Zhaoyang Liu, Linglin Jing, Shenglong Ye, Jie Shao, and 1 others. 2025. Internvl3.5: Advancing open-source multimodal models in versatility, reasoning, and efficiency. arXiv preprint arXiv:2508.18265.
  30. Longhui Wei, Shiliang Zhang, Wen Gao, and Qi Tian. 2018. Person transfer gan to bridge domain gap for person re-identification. In CVPR, pages 79–88. find
  31. Zhengyuan Yang, Linjie Li, Kevin Lin, Jianfeng Wang, Chung-Ching Lin, Zicheng Liu, and Lijuan Wang. 2023. The dawn of lmms: Preliminary explorations with gpt-4v (ision). arXiv preprint arXiv:2309.17421, 9(1):1.
  32. Haiyang Yu, Jinghui Lu, Yanjie Wang, Yang Li, Han Wang, Can Huang, and Bin Li. 2025a. Eve: Towards end-to-end video subtitle extraction with vision-language models. arXiv preprint arXiv:2503.04058. — [arXiv:2503.04058](https://arxiv.org/abs/2503.04058) find
  33. Haiyang Yu, Siyang Yi, Ke Niu, Minghan Zhuo, and Bin Li. 2025b. Umit: Unifying medical imaging tasks via vision-language models. arXiv preprint arXiv:2503.15892. — [arXiv:2503.15892](https://arxiv.org/abs/2503.15892) find
  34. Ye Yuan, Wuyang Chen, Yang Yang, and Zhangyang Wang. 2020. In defense of the triplet loss again: Learning robust person re-identification with fast approximated triplet loss and label distillation. In CVPR Workshops, pages 354–355. find
  35. Guiwei Zhang, Zhang Yongfei, Zhang Tianyu, and Shiliang Pu Bo Li1. 2023. Pha: Patch-wise high-frequency augmentation for transformer-based person re-identification. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 14133–14142. find
  36. Zhizheng Zhang, Cuiling Lan, Wenjun Zeng, Xin Jin, and Zhibo Chen. 2020. Relation-aware global attention for person re-identification. In Proceedings of the ieee/cvf conference on computer vision and pattern recognition, pages 3186–3195. find
  37. Liang Zheng, Liyue Shen, Lu Tian, Shengjin Wang, Jingdong Wang, and Qi Tian. 2015. Scalable person re-identification: A benchmark. In ICCV, pages 1116–1124. find
  38. Liang Zheng, Hengheng Zhang, Shaoyan Sun, Manmohan Chandraker, Yi Yang, and Qi Tian. 2017. Person re-identification in the wild. In CVPR, pages 1367– 1376. find
  39. Kuan Zhu, Haiyun Guo, Tianyi Yan, Yousong Zhu, Jinqiao Wang, and Ming Tang. 2022. Pass: Part-aware self-supervised pre-training for person re-identification. In European conference on computer vision, pages 198–214. Springer. find
  40. Zequn Xie, Chuxin Wang, Yeqiang Wang, Sihang Cai, Shulei Wang, and Tao Jin. 2025. Chat-driven text generation and interaction for person retrieval. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 5259–5270.
  41. Qize Yang, Ancong Wu, and Wei-Shi Zheng. 2020. Person re-identification by contour sketch under moderate clothing change. IEEE Transactions on Pattern Analysis and Machine Intelligence (DOI 10.1109/TPAMI.2019.2960509). — [doi:10.1109/tpami.2019.2960509](https://doi.org/10.1109/tpami.2019.2960509) find
  42. Shan Yang and Yongfei Zhang. 2024. Mllmreid: Multimodal large language model-based person re-identification. arXiv preprint arXiv:2401.13201. — [arXiv:2401.13201](https://arxiv.org/abs/2401.13201) find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/luo-2026-iterative-self-correction-text-driven/