Cross-domain pedestrian re-identification method and system based on three stages

Abstract

The invention discloses a cross-domain pedestrian re-identification method and system based on three stages, and the method comprises the steps: a domain adaptive learning stage: carrying out the processing of a source domain image and a target domain image through a domain adaptive network, calculating each loss, and updating the parameters of the domain adaptive network; a self-supervised training stage: carrying out the supervised training on the domain self-adaptive network through pseudo tags, calculating the loss of the triad difficult to sample, and updating network parameters; a jointloss training stage: constructing a joint loss training network, and defining label smooth regularization loss and difficult-to-sample triple loss; for a target domain image, inputting a joint loss training network, calculating the losses and updating the parameters of the joint loss training network, and carrying out the re-identification of cross-domain pedestrians. According to the cross-domainpedestrian re-identification method, the domain self-adaptation stage, the self-supervision clustering re-training stage and the joint loss learning stage are effectively integrated, and compared with a single training mode, the cross-domain pedestrian re-identification accuracy is further improved. The cross-domain pedestrian re-identification method has the advantages that the cross-domain self-adaptation stage, the self-supervision clustering re-training stage and the joint loss learning stage are effectively integrated.

Description

technical field [0001] The invention belongs to the technical field of cross-domain pedestrian re-identification, and in particular relates to a three-stage cross-domain pedestrian re-identification method and system. Background technique [0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art. [0003] In recent years, with the increase of pedestrian images generated by the camera monitoring system, pedestrian re-identification technology has been increasingly applied to various scenarios, such as criminal investigation and tracking, crime prevention, traffic control, and finding missing persons. [0004] In the field of supervised person re-identification, the feature extraction method has developed from manual extraction to convolutional network extraction, and has greatly improved the retrieval accuracy. However, due to the expensive production cost of the dataset, a large...

AI Technical Summary

Problems solved by technology

[0006] The inventor found in the research that with the emergence of various cross-domain person re-identification models, the accuracy of cross-domain person re-identification has been improved, but most of the existing methods only use one training method, and have not effectively integrated multiple training methods to further improve effect, and the cross-domain training model only considers the differences between different data sets, while ignoring the differences between different shot data in the target domain, thus reducing the model's cross-shot discrimination ability

At the same time, due to the high error rate of pseudo-labels, the effect of representation learning and metric learning on target domain data is not high, so the accuracy of existing cross-domain person re-identification is still at a low level, and the model cannot learn in the target domain. Satisfactory generalization ability and high discriminative features

Images