Confidence-guided outlier refinement and collaborative embedding for unsupervised person re-identification
Jun Zhang, Shuli Cheng, Anyu Du, Haixu Yang · 2026
Read Confidence-guided outlier refinement and collaborative embedding for unsupervised person re-identification by Jun Zhang, Shuli Cheng, Anyu Du, Haixu Yang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13550569.txt.
Description
Jun Zhang, with Shuli Cheng, Anyu Du, and Haixu Yang, published a 2026 computer vision study on matching the same person across different cameras without labeled training data. Unsupervised methods must invent their own labels by clustering, yet stray images and busy backgrounds make those labels noisy. The authors propose cleaning up doubtful cases while also learning richer image features.
One part grades each outlier by its distance to nearby examples and lets neighbors vote to fix the labels of low-confidence images. A second part trains global and local features together, so broad category differences and fine details support each other. Tests on standard data sets showed better clustering and cleaner labels, with steadier performance in complex scenes.
Catalog record
| CREATOR | Jun Zhang, Shuli Cheng, Anyu Du, Haixu Yang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 129470 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |