Abnormity detection method based on generative adversarial network

Abstract

The invention discloses an abnormity detection method based on a generative adversarial network. The abnormity detection method comprises the steps that S1: a semantic segmentation image data set is processed so as to obtain a data set meeting the original image and a data set of Mask images; S2: an abnormity detection generative adversarial network structure including a generative network and a discrimination network is established; S3: the data set meeting the original image and the data set of the Mask images are trained so as to obtain a generative adversarial network model of abnormity detection; and S4: the original image is inputted and the image of the specific abnormity type is obtained. Compared with the methods in the prior art, the abnormity detection method based on the generative adversarial network has the following advantages: 1. the abnormity part image acts as the objective of model training, and the abnormity part concrete locating information acts as the training objective in comparison with the conventional abnormity detection method so that the abnormity part of the image is enabled to be further visual; and 2. compared with the single integrated network structure of the conventional abnormity detection method, the generative adversarial network model is established so that the disadvantage that the image of the abnormity part cannot be outputted can be compensated.

Description

technical field [0001] The invention belongs to the field of GAN image processing, and mainly relates to street outdoor anomaly detection, specifically, a street anomaly detection method based on production confrontation network. Background technique [0002] Generative confrontation network GAN is a kind of generative confrontation model, influenced by game theory, the model usually consists of a generator and a discriminator. The generator captures the underlying distribution of real data and generates new data samples; the discriminator is a binary classifier that identifies whether the input data is real data or a sample generated by the generator. Traditional generative adversarial networks aim to capture the real data distribution from random Gaussian noise, with the purpose of generating pictures that are sufficiently fake. Both the generator and discriminator network structures are convolutional neural networks. [0003] Generative confrontation network involves th...

AI Technical Summary

Problems solved by technology

The idea of ​​generating confrontation makes up for the shortcomings of the inability to generate abnormal Mask areas in anomaly detection

Images