Surface defect detection method based on multi-scale convolution and trilinear global attention

Abstract

The invention discloses a surface defect detection method based on multi-scale convolution and trilinear global attention, and the method comprises the steps: carrying out the convolution and poolingof trunk features in an encoding module, and extracting shallow feature maps of an image under different scales; obtaining a deep feature map through up-sampling and convolution operations in a decoding module; fusing the shallow feature maps and the deep feature map together through four times of splicing operation in the middle; converting the shallow feature map into a shallow attention map bya first branch of the trilinear global attention module through linear operation, activating the deep feature map by a second branch through compression to obtain a deep feature weight of the deep feature map, and then weighting the deep feature weight to the shallow feature map; in a decision-making network module, processing an output feature map of the decoding module by using global average pooling and global maximum pooling, outputting the probability that a surface defect image has defects through an activation function, and outputting a grey-scale map of a potential position of the defect through 1*1 convolution operation for visually explaining a neural network.

Description

technical field [0001] The present invention relates to the field of industrial defect detection, in particular to a surface defect detection method based on multi-scale convolution and trilinear global attention, which can significantly reduce the cost of sample labeling without reducing the performance of the model, and is committed to eliminating weak The effective position information and shape information of defective objects are obtained in the supervised learning label. Background technique [0002] In the process of industrialization, due to irresistible factors, defects may appear on the surface of industrial devices. Therefore, surface defect detection is a basic task to ensure production quality. Usually, manual screening is required, and workers need to spend a lot of time on training. However, this method is inefficient, highly subjective, and limits the production efficiency of products. Usually, the traditional defect detection method follows the same process...

AI Technical Summary

Problems solved by technology

[0004] However, most existing deep learning methods have certain limitations: First, most deep learning methods are data-driven techniques that often require a large amount of labeled sample data to play a role

Third, in the existing defect detection, only the deepest features are used for prediction, and shallow features are not fused; fourth, the existing defect detection is often experimented on cloth images with high repeatability of the texture background, Experiments were not carried out in actual industrial images, thus ignoring the problems of large defect size differences in industrial images and complex industrial environment background with small defects. Therefore, in industrial applications, it is often unable to meet the needs of practical applications.

Images