Automatic bridge crack detections: A lightweight segmentation model for UAV inspection
Xuwei Dong, Zhibing Yu, Jiashuo Yuan, Jinpeng Dai · 2026
Read Automatic bridge crack detections: A lightweight segmentation model for UAV inspection by Xuwei Dong, Zhibing Yu, Jiashuo Yuan, Jinpeng Dai (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-0358865.txt.
Description
Xuwei Dong and three colleagues build a featherweight vision model for finding cracks in bridges, in PLOS ONE in 2026. Cracks threaten a bridge's safety and durability, but sending people to inspect by hand is slow and inconsistent, while heavy software strains the drones that could do the job. Their network aims for precision that still runs lightly enough for aerial inspection gear.
Two new parts do the work: one fuses wide scene meaning with fine detail, and another repairs thin crack features cheaply where image layers join. On a public concrete dataset the model marked crack pixels with accuracy above eighty percent on the strict overlap measure, beating several established networks while staying small. A companion detector handles messy real world backgrounds, moving the system closer to routine drone surveys of aging bridges.
Catalog record
| CREATOR | Xuwei Dong, Zhibing Yu, Jiashuo Yuan, Jinpeng Dai |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51637 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |