Efficient Vision Language Model Fine-tuning for Text-based Person Anomaly Search DOI: 10.1145/XXXXXXX Conference: Sydney ’25: ACM Web Conference; 28 April-2 May 2025; Sydney, Australia Woodstock ’18: ACM Symposium on Neural Gaze Detection, June 03–05, 2018, Woodstock, NY Price: 15.00 ISBN: 978-1-450
Jiayi He Shengeng Tang Ao Liu Lechao Cheng Jingjing Wu Yanyan Wei · 2025
Read Efficient Vision Language Model Fine-tuning for Text-based Person Anomaly Search DOI: 10.1145/XXXXXXX Conference: Sydney ’25: ACM Web Conference; 28 April-2 May 2025; Sydney, Australia Woodstock ’18: ACM Symposium on Neural Gaze Detection, June 03–05, 2018, Woodstock, NY Price: 15.00 ISBN: 978-1-450 by Jiayi He Shengeng Tang Ao Liu Lechao Cheng Jingjing Wu Yanyan Wei (2025) 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/arxiv-2502-03230.txt.
Description
Jiayi He and co-authors present their winning approach to a 2025 web conference challenge on finding anomalous pedestrians from text descriptions, in a 2025 paper. The task pairs written descriptions with a huge library of pedestrian images, and the hard part is telling apart people whose descriptions sound nearly alike. The team fine-tuned a vision-language model for the search.
Their Similarity Coverage Analysis strategy targets exactly that confusion, teaching the model to weigh the small details that separate similar descriptions. The authors report that the approach sharply improved both accuracy and reliability of matches. The paper frames the method as a general lesson for fine-grained search where near-duplicate text is the norm.
Catalog record
| CREATOR | Jiayi He Shengeng Tang Ao Liu Lechao Cheng Jingjing Wu Yanyan Wei |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 20 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-03230