An Algorithm for Real-Time Aluminum Profile Surface Defects Detection Based on Lightweight Network Structure

Surface defects, which often occur during the production of aluminum profiles, can directly affect quality and should be monitored in real time. This paper proposes an effective, lightweight detection method for profiles to realize real-time surface defect with ensured accuracy. Based on YOLOv5s framework, a network model is designed by adding attention mechanism depth-separable convolution aluminum. The improves limitations framework regarding its accuracy speed. backbone GCANet built based Ghost module, Attention module embedded AC3Ghost module. A compression achieved, more channel information focused on. size further reduced compressing Neck using deep separable convolution. experimental results show that, compared YOLOv5s, proposed mAP 1.76%, reduces 52.08%, increases speed factor two. Furthermore, reach 17.4 FPS Nvidia Jeston Nano’s edge test, achieves detection. It also provides possibility embedding devices industrial inspection.