基于机器视觉的显示屏表面缺陷检测方法

Defect Detection Methods for Screen Based on Machine Vision

传统显示屏表面缺陷检测方法存在算法设计成本高、鲁棒性差的缺点,难以适应工业4.0的时代背景。为提高显示屏表面缺陷检测效率,提出了一种改进轻量化神经网络模型的方法。首先,引入GhostNet轻量化卷积神经网络模型,提升检测速度;其次,引入CA注意力机制改进网络,增强模型对显示屏表面轻微、细小疵点的检测精度。实验结果显示,在显示屏表面缺陷数据集上,所提方法的检测精度达到96.7%,在基线网络的基础上提升了4.0%,FPS达到了124,实现了快速高精度检测。

The traditional defect detection methods for screen have the disadvantages of high algorithm design cost and poor robustness,which is difficult to adapt to the era of Industry 4.0.In order to improve the efficiency of screen defect detection,this paper proposes an improved lightweight neural network model method.Firstly,the GhostNet model is introduced to improve the detection speed.Secondly,the CA attention mechanism is introduced to improve the detection accuracy of the model for minor and small defects on the screen.The experimental results show that on the screen defect dataset,the detection accuracy of our proposed model reaches 96.7%,which is 4.0%higher than that of the baseline network,and the FPS reaches 124,which realizes fast and high-precision detection.