Abnormal image detection method based on a supervised generative adversarial network

Abstract

The invention relates to an abnormal image detection method based on a supervised generative adversarial network. The method comprises the following steps: 1) obtaining a to-be-detected picture; 2) inputting the to-be-detected picture into a trained Supervised GANomal model to obtain a corresponding abnormal category trust value; 3) judging whether the abnormal category trust value is greater thanan abnormal category discrimination threshold, if yes, judging the abnormal image as an abnormal image, and if not, judging the abnormal image as a normal image; Wherein the Supervised GANomain modelcomprises a generation network, a coding network, a discrimination network and a classification network, the generation network, the coding network and the discrimination network are used for learning feature distribution of a normal sample, and the classification network is used for distinguishing the normal sample from an abnormal sample. Compared with the prior art, the method has the advantages that the normal sample and the abnormal sample can be effectively distinguished, the robustness is good and the like.

Description

technical field [0001] The invention relates to the technical field of abnormal image detection, in particular to an abnormal image detection method based on a supervised generative confrontation network. Background technique [0002] With the continuous development and popularization of artificial intelligence methods, automatic detection technology has received more and more attention. Abnormal image detection, as an important branch of automated detection, has played an increasingly important role in industrial quality inspection, medical diagnosis, automatic driving, security and other fields. [0003] Abnormal image detection methods can be roughly divided into three categories, supervised methods, semi-supervised methods and unsupervised methods. [0004] Supervised anomaly detection methods are similar to classification methods in machine learning, but in anomaly detection problems, there are usually more normal samples, while abnormal samples are usually very few, a...

AI Technical Summary

Problems solved by technology

[0008] The purpose of the present invention is to provide an abnormal image detection method based on a supervised generation confrontation network in order to overcome the defects of the above-mentioned prior art, so as to solve the problem that the existing generation confrontation network-based method cannot use supervision information and cannot detect the difference between appearance and normal samples The problem with relatively small outlier samples

Images