Video pedestrian re-identification algorithm and device based on multi-space-time attention model

Abstract

The invention discloses a video pedestrian re-identification algorithm and device based on a multi-space-time attention model. The algorithm comprises: firstly, using a multi-space attention model forpositioning and distinguishing a video image pedestrian area; calculating the aggregate representation of the features extracted by each spatial attention model in combination with the time attentionmodel, and carrying out pedestrian re-identification loss function compensation; and finally comparing the feature vector of the to-be-identified image with the feature vector of the pedestrian of the video image in the effective geographic area range, searching the pedestrian target image with the highest similarity, and outputting a final re-identification matching result. The algorithm can solve the problems that in the actual process, the pedestrian video image imaging quality is poor, detail display is not clear, useful information of a pedestrian part is blocked, feature extraction is difficult, and consequently the video monitoring system is not high in re-identification accuracy, and the algorithm meets the real-time requirement.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a video pedestrian re-identification algorithm and device based on a multi-temporal-spatial attention model. Background technique [0002] With the continuous development of artificial intelligence, computer vision and hardware technology, video image processing technology has been widely used in smart city systems. [0003] Person re-identification (Person Re-identification) is also called person re-identification, or Re-ID for short. It is a technology that uses computer vision technology to judge whether there is a specific pedestrian in an image or video sequence. In surveillance video, due to camera resolution and shooting angle, it is usually impossible to obtain very high-quality face pictures. When face recognition fails, Re-ID becomes a very important substitute technology. A very important feature of Re-ID is cross-camera, so retrieving the same pedestrian pictures und...

AI Technical Summary

Problems solved by technology

In practice, pedestrians are often partially occluded, which destroys the extracted features

And video-based pedestrian re-identification Re-ID, if all the frames are used, there will be a lot of poor quality pictures (occlusion, lighting and other factors) that will have a great impact on the features of deep learning

Thus affecting the accuracy of pedestrian re-identification, increasing the amount of calculation of the system and deteriorating the robustness of the system

Images