A bridge crack detection method fusing dynamic features and global context
By improving the backbone network and feature fusion module, the problems of lack of subtle feature representation and global context information in bridge crack detection are solved, realizing efficient, accurate and automated identification and intelligent evaluation of bridge cracks, thus improving detection efficiency and accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ANQING VOCATIONAL & TECHN COLLEGE
- Filing Date
- 2026-03-25
- Publication Date
- 2026-07-17
AI Technical Summary
Existing bridge crack detection technologies suffer from insufficient ability to represent subtle features, poor fusion of multi-scale features, and a lack of global contextual information, resulting in high rates of missed and false detections, making it difficult to meet the requirements of modern transportation systems for detection speed and accuracy.
An improved backbone network is adopted, which embeds an attention-enhanced C2f module that enhances the local and global modules, and combines a multi-level feature modulation module and a content-guided attention fusion module to achieve adaptive weighted fusion of multi-scale features and precise complementarity of low-level details and high-level semantic features, and outputs the bounding box, confidence and category information of the crack target.
It has achieved efficient, accurate and automated identification of bridge cracks, automatically extracted crack geometric parameters, and carried out damage level assessment and intelligent diagnosis of causes, forming a fully intelligent detection and evaluation system, which improves detection efficiency and engineering practical value.
Smart Images
Figure CN122415973A_ABST