Yolov8-CA-EYHL: A bridge crack detection algorithm based on improved YOLOv8
Xianwei Zhu, He Chao, Yahui Zhang · 2026
Read Yolov8-CA-EYHL: A bridge crack detection algorithm based on improved YOLOv8 by Xianwei Zhu, He Chao, Yahui Zhang (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-0358872.txt.
Description
Xianwei Zhu, He Chao, and Yahui Zhang upgrade a popular detection network to find cracks in bridge concrete, in PLOS ONE in 2026. Cracks are thin, faint, and irregular, and busy surface textures fool automated eyes. Their version of the YOLOv8 detector first cleans the image with filtering and contrast equalization, then adds attention modules that respect direction and mix fine and broad features across scales.
The redesign jumped detection quality dramatically over the stock network on their benchmark, while keeping the model small enough to run at roughly seventy-six frames per second. Step by step removal tests showed each added part contributed, rather than one trick doing all the work. Tests with different random seeds and on an independent outside dataset held up, supporting the claim that the system generalizes. For inspection teams, it promises fast first pass scans that flag slender cracks human reviewers can then verify.
Catalog record
| CREATOR | Xianwei Zhu, He Chao, Yahui Zhang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45582 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |