Pedestrian re-identification method and system based on momentum network and comparative learning

Abstract

The invention discloses a pedestrian re-identification method and system based on a momentum network and comparative learning. The method comprises the following steps: obtaining a pedestrian picture and carrying out labeling processing; initializing a momentum network and a backbone network; pedestrian picture features are extracted; carrying out comparison loss calculation on the average feature set of the pedestrian pictures and updating a backbone network and a momentum network; hierarchical clustering processing is carried out on the average feature set of the pedestrian pictures, and pseudo labels are given; carrying out comparison loss calculation on a clustering result and updating a backbone network and a momentum network; updating the pedestrian picture average feature set; and iteratively optimizing the pedestrian re-identification model for multiple times, and outputting the optimized pedestrian re-identification model. According to the method, the interference of noise in the data training process is reduced by designing the momentum network, and the pedestrian re-recognition effect can be improved under the condition that the manual labeling cost is reduced. The pedestrian re-identification method and system based on the momentum network and comparative learning can be widely applied to the technical field of computer vision.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a pedestrian re-identification method and system based on momentum network and contrastive learning. Background technique [0002] With the development of deep learning, neural network technology is applied in more and more scenarios, and person re-identification, as a popular research direction in the field of computer vision, is also getting more and more attention. Person re-identification (Re-ID) is mainly To solve the problem of retrieving and matching the same pedestrian under different cameras, the final retrieval result is obtained by comparing and sorting the pedestrian image features to be queried with the pedestrian image features in the retrieval database. Pedestrian re-identification has a wide range of application scenarios and great practical significance in smart cities, security and other fields. In the process of marking pedestrian pictures, it is relatively easy ...

AI Technical Summary

Problems solved by technology

[0003] In order to solve the above technical problems, the purpose of the present invention is to provide a pedestrian re-identification method and system based on momentum network and contrastive learning, which can improve the pedestrian re-identification effect while reducing the cost of manual labeling

Images