Pipe crack visual detection method based on simi-attention and boundary sensitive loss function
By employing a collaborative mechanism of Simi-Attention and boundary-sensitive loss function, the performance-efficiency contradiction of attention mechanisms and the insufficient accuracy of bounding box localization in pipeline crack detection are resolved. This enables lightweight and efficient pipeline crack detection, adapting to complex environments, reducing equipment costs, and improving detection accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HENAN POLYTECHNIC UNIV
- Filing Date
- 2026-04-22
- Publication Date
- 2026-07-17
AI Technical Summary
Existing pipeline crack detection technologies suffer from performance-efficiency contradictions in attention mechanisms, insufficient bounding box positioning accuracy, and poor adaptability to complex scenarios, making it difficult to meet the requirements of lightweight, real-time inspection, and high-precision detection.
A visual detection method for pipe cracks based on Simi-Attention and boundary-sensitive loss function is adopted. Feature extraction and fusion are performed through a lightweight target detection model, a parameterless SimAM attention module, a feature pyramid network and a path aggregation network. The model is trained and optimized by combining the boundary-sensitive loss function and deployed to an embedded edge computing unit for real-time detection.
It achieves a balance between lightweight design and high efficiency, improves boundary positioning accuracy, adapts to complex environments, meets real-time detection requirements, reduces equipment deployment costs, and improves operation and maintenance efficiency and detection accuracy.
Smart Images
Figure CN122415518A_ABST