基于压缩激励模块的工业产品缺陷分类算法

Defect Classification Algorithm for Industrial Products Based on Squeeze Excitation Module

针对传统的工业产品表面缺陷检测技术检测效率低下、检测速度慢等问题,在以前工作的基础上,提出了基于压缩激励模块的工业产品缺陷分类算法。主要是在原有的并行模块与并行非对称卷积模块基础上增加了压缩和激励机制,通过显式的建模特征通道之间的相互依赖关系来提升网络的表征能力。在偏光片图像数据集上的实验结果表明,在不大量增加模型占用内存大小以及单张图片测试时间的基础上,显著提高了缺陷分类的正确率。

Aiming at the problems of low detection efficiency and slow detection speed of traditional surface defect detection technology of industrial products,on the basis of previous work,a defect classification algorithm for industrial products based on squeeze excitation module was proposed.Compression and incentive mechanisms are added on the basis of the original parallel module and parallel asymmetric convolution module to improve the representation ability of the network by explicitly modeling the interdependence between feature channels.The experimental results on polarizer dataset show that the accuracy of defect classification is significantly improved without greatly increasing the size of model and the test time per image.