Vision expert guided inspection for industrial anomaly detection
Xiangyu Zhu, Wenhua Cui, Ye Tao, Xilong Wang · 2026
Read Vision expert guided inspection for industrial anomaly detection by Xiangyu Zhu, Wenhua Cui, Ye Tao, Xilong Wang (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/plos-pone-0353291.txt.
Description
Xiangyu Zhu and three colleagues present this 2026 PLOS One paper on spotting defects in manufactured products with computer vision. Factory inspection systems often miss faint, subtle flaws, and training data for rare defects is scarce. Their method first sharpens suspicious image regions with super-resolution, then combines information across image scales using a graph-based network that weighs how regions relate to each other.
Tested on public benchmark datasets with only a few example images, or none at all, the method beat leading alternatives at finding defects and pinpointing their exact location. It performed especially well on fine textures and held up when moved to new domains. The authors present it as a practical step toward inspection systems that catch the flaws human eyes and older models overlook.
Catalog record
| CREATOR | Xiangyu Zhu, Wenhua Cui, Ye Tao, Xilong Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58523 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |