车辆表面缺陷检测方法、装置、计算机设备及存储介质
By using the GS-YOLO enhanced lightweight model and integrating multi-module collaborative optimization technology, the shortcomings of vehicle surface defect detection models in terms of irregular defect modeling, small target detection, balance between accuracy and efficiency, and robustness are solved, thus achieving high-precision and high-efficiency vehicle surface defect detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGDONG POLYTECHNIC NORMAL UNIV
- Filing Date
- 2026-03-17
- Publication Date
- 2026-07-17
AI Technical Summary
Existing vehicle surface defect detection models have shortcomings in terms of irregular defect modeling capabilities, small target detection, balance between accuracy and efficiency, bounding box regression stability, and multi-module collaboration, making it difficult to achieve high-precision and high-efficiency detection in complex scenarios.
We adopt the GS-YOLO enhanced lightweight model, which integrates a gated deformable attention module, a BiFPN-D enhanced feature fusion module, and a C2f_GS lightweight feature extraction module. Through multi-module collaborative optimization, combined with the WIoU loss function and data augmentation techniques, we improve the model's geometric modeling ability, feature extraction ability, and robustness.
It significantly improves the detection accuracy of irregular defects and the detection rate of small targets, reduces computational complexity, enhances the robustness and generalization ability of the model in complex scenarios, and is suitable for real-time deployment on edge devices.
Smart Images
Figure CN122415441A_ABST