Defect Detection Methods for Screen Based on Machine Vision
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.
defect detection for screenvisual saliency methodsneural network
NETL
NSTL
万方数据
