Method, system and device for constructing few-sample industrial image defect detection model

Abstract

The invention discloses a few-sample industrial image defect detection model construction method, system and device. The method comprises the following steps: dividing an industrial image data set into a training set and a test set; preprocessing the training set and the test set to enhance the image contrast; performing random vertical overturning and horizontal overturning on the preprocessed training set to obtain training sample data; inputting training sample data into the segmentation network for training, and completing first-step construction of a detection model; correcting the mask image to obtain an industrial image defect area, and completing second-step construction of the detection model; splicing the mask pattern and the industrial image defect area to obtain a dual-channel feature pattern, and inputting the dual-channel feature pattern into a decision network for training to complete third-step construction of a detection model; and evaluating the detection model according to the defect classification result of the industrial images of different defect types in the test set. According to the invention, the detection precision and speed of the defect image can be improved, and the constructed model is evaluated.

Description

technical field [0001] The invention relates to the field of constructing a few-sample industrial image defect detection model, in particular to a method, system and device for constructing a few-sample industrial image defect detection model. Background technique [0002] At present, the methods of industrial image defect detection can be divided into two categories: methods based on traditional machine vision and methods based on deep learning. [0003] The traditional machine vision method is mainly based on the use of threshold segmentation, morphological processing, wavelet transform, edge detection and other algorithms to achieve defect detection. In deep learning, convolutional neural networks are mainly used to extract features from images. For example, classification networks, segmentation networks, and target detection networks are used to achieve defect detection. The commonly used models are: ResNet, YOLO, U-Net, etc. [0004] The method based on traditional mac...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a method, system and device for building a few-sample industrial image defect detection model, aiming at solving the problem of building a few-sample industrial image defect detection model

Images