Deep learning defect image recognition method and system based on ensemble learning

Abstract

The invention provides a deep learning defect image recognition method and system based on ensemble learning. The scheme comprises the following steps: acquiring all defect monitoring images, and generating a sample training set and a sample test set; obtaining a sample training set, obtaining a grey-scale map, a spectrogram, an edge contour map and a gradient map through color conversion, Fourier transform, gradient operation and edge contour extraction, generating a first training set, a second training set, a third training set and a fourth training set, and fusing to generate a fifth training set; performing deep neural network training on the first training set, the second training set, the third training set, the fourth training set and the fifth training set to generate a first classifier, a second classifier, a third classifier, a fourth classifier and a fifth classifier; and voting the images in the sample test set to obtain a target classification result. According to the scheme, the defect recognition universality of the network model is improved through a multi-training-set and integrated learning mode, and multi-class defect image recognition is achieved.

Description

technical field [0001] The present invention relates to the technical field of computer vision image recognition, and more specifically, to a deep learning defect image recognition method and system based on integrated learning. Background technique [0002] The purpose of defective image detection is to identify defective images in sample files and mark them. Defect image detection has always been a difficult problem in industrial vision inspection. When artificially extracting the features of defective images, it is difficult to achieve accurate and complete results. Therefore, with the continuous development of deep learning technology in recent years. More and more fields have begun to consider the use of deep learning methods to recognize and classify defect images. Compared with the traditional artificial defect detection method, its effect has been significantly improved. [0003] However, before the technology of the present invention, the traditional image defec...

AI Technical Summary

Problems solved by technology

[0003] However, before the technology of the present invention, the traditional image defect recognition model based on deep learning has the problem of poor versatility

Therefore, when it is necessary to identify multiple types of different defects, there are often situations such as missed detection or detection failure.

Images