Pedestrian re-identification method and system fused with part texture three-dimensional mapping, medium and terminal

Abstract

The invention provides a pedestrian re-identification method and system fused with part texture three-dimensional mapping, a medium and a terminal, and the method comprises the following steps: extracting the pedestrian texture information of a pedestrian image, and obtaining a pedestrian texture map containing the pedestrian texture information; preprocessing the pedestrian picture and the pedestrian texture map; and training a pedestrian re-identification network model based on the pre-processed pedestrian image and the pedestrian texture map so as to perform pedestrian re-identification based on the trained pedestrian re-identification network model. By constructing the pedestrian re-identification feature extraction method with attitude robustness and the deep learning network taking the features as auxiliary data, the change of pedestrian attitude and environment can be effectively coped with, an attention mechanism is combined, the information fusion under different spatial features is realized, and the pedestrian re-identification efficiency and accuracy are improved.

Description

technical field [0001] The present invention relates to the technical field of pedestrian re-identification, in particular to a pedestrian re-identification method, system, medium and terminal for integrating three-dimensional mapping of part textures. Background technique [0002] Person Re-Identification (Person Re-Identification, also referred to as ReID) is one of the computer vision application environments that have been widely discussed in recent years. Analyzing pedestrian behavior rules, this technology has good prospects in business analysis, criminal tracking, etc., but the current research results have not yet reached industrial standards. [0003] On the premise that there is a dataset of framed pedestrians, the current research on improving the efficiency of pedestrian re-identification algorithms is mainly divided into the following categories: [0004] (1) Application of attention mechanism: In addition to obtaining global features, this type of method inclu...

AI Technical Summary

Problems solved by technology

[0005] (2) Environment or pose migration: The effectiveness of the pedestrian re-identification algorithm in different data sets is quite different, because the acquisition equipment and lighting environment between different data sets are quite different; the training set on the same data set The results of the test set and the test set may also have a large difference. In addition to the possibility of overfitting, it is more of a pedestrian posture, and the diversity of angles leads to

[0007] Although many methods have been invented to improve the efficiency of ReID, the defects of various algorithms are also very obvious: violent segmentation of pedestrian pictures can obtain the highest accuracy on the current data set, but the real camera is not level For pedestrians, this leads to the fact that pedestrians at different distances may have different body proportions under the camera. This change may lead to a significant decline in the effect of the model in actual application scenarios; posture and environment migration are relatively time-consuming, especially in the amount of data. In the case of a large number of shots; as the pedestrian database continues to expand, there is no good way to compress the amount of data for the time being, because changes in posture and environment make the same pedestrian picture have a lot of changes

Images