A welding seam defect detection method and system based on an improved YOLO model
By improving the Aura-YOLO model of YOLO and adopting the SPDConv, HFPN and AoV-GSCSP modules, combined with the GFL loss function, the problems of small target feature loss and insufficient fusion of complex background in weld defect detection are solved, and high-precision, real-time weld defect detection is achieved.
CN122434840APending Publication Date: 2026-07-21HANGZHOU DIANZI UNIV +1
View PDF 0 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HANGZHOU DIANZI UNIV
- Filing Date
- 2026-04-15
- Publication Date
- 2026-07-21
Smart Images
Figure CN122434840A_ABST
Abstract
The application provides a kind of welding seam defect detection method and system based on improved YOLO model, the method first obtains welding seam X-ray image and carries out pretreatment, subsequently constructs improved YOLO model, wherein the improved YOLO model includes including backbone network, neck network and detection head loss function;The backbone network adopts space-depth conversion convolution module as down-sampling layer, the neck network includes light weight feature fusion module and hierarchical feature fusion module, the detection head is used to predict the class, position and confidence of defect according to the feature map output by neck network;Finally, the welding seam X-ray image is input into improved YOLO model, and the detection result of welding seam defect is obtained.The present application significantly improves the detection rate of micro-defects, enhances the recognition accuracy in complex background, realizes the optimal balance of high precision and real-time, and can effectively solve the problem of small target missed detection and weak contrast defect recognition in welding seam detection.
Need to check novelty before this filing date? Find Prior Art