基于YOLOv7的工件表面缺陷实时检测系统研究

Research on Real-time Detection System of Workpiece Surface Defects Based on YOLOv7

针对现有工件表面缺陷检测方法在准确率、实时性及效率方面的不足,提出一种基于YOLOv7深度学习算法的工件表面缺陷检测系统模型。该模型在保证性能的同时扩大了检测范围,优化了模型结构,解决了作为工件表面缺陷检测主要难点之一的小目标缺陷检测。试验结果表明,与改进前相比,在滚珠丝杠表面缺陷的检测中该模型的精确率得到了明显提高,检测速度和精度均达到实际工业生产效率需求。

Aiming at the shortcomings of the existing workpiece surface defect detection methods in accuracy,real-time and efficiency,this paper proposes a workpiece surface defect detection system model based on YOLOv7 deep learning algorithm.The model expands the detection range while ensuring the performance,optimizes the model structure,and solves the small target defect detection,which is one of the main difficulties of workpiece surface defect detection.The experimental results show that the accuracy of the model is significantly improved in the detection of surface defects of ball screw,and the detection speed and accuracy meet the requirements of actual industrial production efficiency.