Lightweight Network-Based Surface Defect Detection Method for Steel Plates

This article proposes a lightweight YOLO-ACG detection algorithm that balances accuracy and speed, which improves on the classification errors missed detections present in existing steel plate defect algorithms. To highlight key elements of desired area surface flaws plates, void space convolutional pyramid pooling model is applied to backbone network. fusion high- low-level semantic information by designing feature networks with embedded spatial attention. According experimental findings, suggested enhances mapped value about 4% once compared YOLOv4-Ghost homemade data set. Additionally, real-time speed reaches 103FPS, 7FPS faster than algorithm, capability defects significantly enhanced meet needs realistic scenes mobile terminal.