Partial pedestrian re-identification method based on visible perception texture semantic alignment

Abstract

The invention discloses a partial pedestrian re-identification method (TSA) based on visible perception texture semantic alignment. The method can simultaneously and efficiently solve the two problemsthat pedestrians are shielded and the posture or observation visual angle is changed. The method comprises the following specific steps of: (1) designing a local area alignment network based on humanbody postures, and mainly solving the problem that pedestrians are shielded; (2) designing a texture alignment network based on semantic visibility, and mainly solving the problem of pedestrian posture change or visual angle change; and (3) in order to enable the model to have better generalization ability, performing joint learning on the two networks so as to enable the model to better deal with the problem that the pedestrians are shielded and the problem that the posture or the observation visual angle is changed. According to the pedestrian re-identification method, efficient partial pedestrian re-identification is carried out based on visible perception texture semantic alignment and human body posture partial region alignment, the problems of shielding and posture change in pedestrian re-identification can be effectively solved, the convergence speed is high, and efficient re-identification can be realized in pedestrian shielding.

Description

technical field [0001] The invention belongs to the technical field of computer vision and pattern recognition, and relates to a partial pedestrian re-identification method (TSA) based on the semantic alignment of visible and perceptual textures, which can simultaneously perform alignment on textures and local areas, and can solve the problem of occlusion and pedestrian re-identification in pedestrian re-identification. The problem of attitude transformation. Background technique [0002] In recent years, with the continuous improvement of infrastructure construction, tens of millions of non-overlapping (disjoint) cameras have been placed in every corner of many cities in order to ensure public security and safety. In some special cases, when a target person disappears from one camera, we want to quickly re-identify the target person from other cameras. This is exactly the research hotspot in the field of computer vision and machine learning - person re-identification (ReID...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to solve the inconsistency of the input image scale in the existing classical algorithms SWM [1] and AWC [1], the problem of non-shared area feature interference in the algorithms of DSR [2] and DCR [3], and the problems not considered in these methods To the problem of human body pose transformation in real scenes, a partial pedestrian re-identification method (TSA) based on visible perceptual texture semantic alignment is provided

Images