A pedestrian re-identification method, system, device, and computer-readable storage medium

Abstract

The present application discloses a pedestrian re-identification method, system, device, and computer-readable storage medium, for obtaining unlabeled first-type pedestrian images; making label information for the first-type pedestrian images; based on the first-type pedestrian images and label information Train the target pedestrian re-identification network to obtain the target pedestrian re-identification network for the first training; extract the area of ​​interest in the first type of pedestrian image; based on the first type of pedestrian image and the area of ​​interest, the target of the first training The pedestrian re-identification network is trained to obtain the target pedestrian re-identification network for the second training, and the pedestrian re-identification is performed based on the target pedestrian re-identification network. The present application can make the pedestrian information in the unlabeled pedestrian image also used for the training of the target pedestrian re-identification network, improve the robustness of the target pedestrian re-identification network, improve the training efficiency of the target pedestrian re-identification network, and further improve the pedestrian re-identification network. Operational efficiency of the identification method; good applicability.

Description

technical field [0001] The present application relates to the technical field of deep learning, and more particularly, to a person re-identification method, system, device and computer-readable storage medium. Background technique [0002] Currently, deep learning is used to solve problems in computer vision fields such as image classification, image segmentation, and object detection with great success. In recent years, many excellent convolutional neural network models have emerged. During this process, the task of person re-identification has also made great progress with the continuous development of deep learning technology. Person re-identification (Re-ID) is an important image recognition technique and is now regarded as a key sub-problem of image retrieval, which utilizes computer vision algorithms to match pedestrian images or videos across devices, that is, given a Query images, and retrieve the same pedestrian from the image database of different monitoring devic...

AI Technical Summary

Problems solved by technology

However, the training process of the existing pedestrian re-identification network is slow, which makes the operation efficiency of the pedestrian re-identification method low, and the robustness of the pedestrian re-identification network is poor, and the applicability of the method is poor.

Images