FS-YOLO: A Lightweight Insulator Defect Detection Method Based on Multi-Level Feature Fusion and Localization Quality Estimation
Congjie Wen, Zhiliang Zhu, Yijian Weng, Zekai Cai, Jiacheng Wang · 2026
Read FS-YOLO: A Lightweight Insulator Defect Detection Method Based on Multi-Level Feature Fusion and Localization Quality Estimation by Congjie Wen, Zhiliang Zhu, Yijian Weng, Zekai Cai, Jiacheng 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/pmc-13611221.txt.
Description
Congjie Wen and four co-authors slimmed down a defect-spotting vision model until it can ride on the drones and pole cameras that inspect power lines. Cracks and damage on insulators are small, the backgrounds are cluttered, and inspection hardware is weak, so the model must be accurate and tiny at once. Their FS-YOLO rebuilds each stage of a standard detector with that tension in mind.
A redesigned backbone extracts features cheaply, a multi-branch neck fuses detail across scales, and a shared detection head grades its own localization quality to claw back accuracy lost to slimming. A final pruning pass cut redundant channels. The finished model holds under a million parameters, yet scores ninety-three percent precision and beats its baseline's accuracy while using about a third less computation. The authors offer it as a reference design for intelligent line inspection that can run at the edge.
Catalog record
| CREATOR | Congjie Wen, Zhiliang Zhu, Yijian Weng, Zekai Cai, Jiacheng Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 64 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |