CNC: Cross-modal Normality Constraint for Unsupervised Multi-class Anomaly Detection
Xiaolei Wang, Xiaoyang Wang, Huihui Bai, Eng Gee Lim, Jimin Xiao · 2025
Read CNC: Cross-modal Normality Constraint for Unsupervised Multi-class Anomaly Detection by Xiaolei Wang, Xiaoyang Wang, Huihui Bai, Eng Gee Lim, Jimin Xiao (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-2501-00346.txt.
Description
Xiaolei Wang and co-authors stop an anomaly spotter from explaining away the very flaws it should catch. Their setting is factory style image checks, with many product classes and no labeled flaws. A common design rebuilds an image and flags where the rebuild fails. Trouble is, a strong rebuilder fixes flaws too, hiding them.
Their fix guides the rebuild with text. Learned prompts capture what normal looks like across classes, in words. The rebuild is pulled toward that normal text sense, so flaws stand out again. A gated expert module further sorts the wild mix of patch types. Scores on two benchmark sets rival the best known.
Catalog record
| CREATOR | Xiaolei Wang, Xiaoyang Wang, Huihui Bai, Eng Gee Lim, Jimin Xiao |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-00346