A Person Re-Identification Method with Illumination Robustness

Abstract

The invention discloses a pedestrian re-identification method with illumination robustness, which comprises the following steps: acquiring a pedestrian re-identification data set with illumination difference, selecting training samples from it, and dividing the training samples into normal illumination pictures and dim illumination pictures ;Initialize a deep convolutional neural network netA, train it with normal lighting pictures; initialize a deep convolutional neural network netB with the same structure as netA, and train it with dim lighting pictures; merge the two kinds of lighting data sets At the same time, netA and netB are jointly trained to make both networks converge; the respective performances of netA and netB are tested, and the network with higher performance is selected as the final model. Without increasing the parameters of the final model, the present invention enhances the recognition ability of the pedestrian re-identification network for dim pictures through the cooperative learning of the double network, and improves the robustness of the model for illumination.

Description

technical field [0001] The invention relates to the fields of image processing and retrieval, in particular to a pedestrian re-identification method with illumination robustness, which solves the problem of poor performance in dark conditions in pedestrian re-identification. Background technique [0002] At present, with the rapid development of computer technology, video surveillance systems are widely used in our daily life and production environment, and play a very important role in maintaining social order. Nowadays, there are a huge number of surveillance cameras in public places, especially in airports, railway stations, supermarkets, highways, etc., which constitute a surveillance network covering most areas, providing security personnel with the ability to accurately locate incident sites and personnel locations. very big help. At present, most of the pedestrian retrieval in surveillance cameras still relies on manpower, requiring operators to play back, record, an...

AI Technical Summary

Problems solved by technology

In actual application scenarios, due to the different lighting environment of each camera, the same pedestrian may have brightness differences in different cameras, which will affect the accuracy of retrieval.

Images