YOLOv8-CAACA: A Context-Aware Adaptive Confidence Adjustment and Target Fusion Algorithm for Pavement Crack Recognition in Complex Scenarios

DOI

10.2991/978-94-6239-648-7_4How to use a DOI?

Keywords

YOLOv8; Road Crack Detection; Context-aware Adaptive Confidence Adjustment; Clustering-Based Merging

Abstract

To address the issues in current crack detection from road images—such as inaccurate identification of long cracks, misdetection and missed detection of cracks in complex environments, and splitting of a single crack into multiple segments—this paper proposes an improved YOLOv8 algorithm (YOLOv8s-CAACA) incorporating context-based adaptive confidence adjustment and crack clustering-merging. The proposed method first integrates a context-based adaptive confidence adjustment module into the YOLOv8s algorithm, enabling the model to more accurately identify road cracks and reduce isolated false detections caused by road potholes, protrusions, or similar conditions. Secondly, a crack clustering module is added to merge cracks within a certain threshold into a single instance, allowing the model to directly recognize entire long cracks without splitting them into fragmented segments that render detection results unreadable. In tests on a dataset of 5,000 images under simple environments, the optimized YOLOv8 model achieved a slight improvement in the evaluation. In tests under complex conditions, however, the optimized algorithm showed significant enhancements and greatly improved the visibility of detection results. Experimental results demonstrate that YOLOv8s-CAACA achieves good accuracy in crack detection under complex environments. Meanwhile, the model maintains lightweight properties, indicating its application value and prospects in road maintenance and upkeep.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Muhan Bai
PY  - 2026
DA  - 2026/04/24
TI  - YOLOv8-CAACA: A Context-Aware Adaptive Confidence Adjustment and Target Fusion Algorithm for Pavement Crack Recognition in Complex Scenarios
BT  - Proceedings of the International Workshop on Advances in Deep Learning for Image Analysis and Computer Vision (IWADIC 2025)
PB  - Atlantis Press
SP  - 20
EP  - 32
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6239-648-7_4
DO  - 10.2991/978-94-6239-648-7_4
ID  - Bai2026
ER  -