A multi-view human behavior recognition method and system based on edge computing architecture

Abstract

The present application provides a multi-view human behavior recognition method and system under an edge computing architecture, which belongs to the technical field of human behavior recognition. The method includes: the camera group shoots the same scene from different angles of view, obtains human behavior video data of different angles of view, and transmits it to the edge computing node connected to it, collects and saves the human behavior video data of different angles of view in the same period of time and performs data processing. Preprocessing, input to the human behavior feature encoder to obtain the multi-view human behavior feature vector, the cloud server receives the multi-view human behavior feature vector uploaded by the edge computing node, and inputs it to the human behavior recognition model to obtain the human behavior recognition result. By extracting human behavior features on the edge computing nodes, the cloud server performs human behavior classification, thereby reducing the computing pressure on the cloud server and improving the real-time performance of recognition; collecting and utilizing multi-view human behavior information, improving the expressive ability of features, and improving human behavior recognition the accuracy rate.

Description

technical field [0001] The present application relates to the technical field of human behavior recognition, in particular to a multi-view human behavior recognition method and system under an edge computing architecture. Background technique [0002] Human behavior recognition technology can judge people's behavior and meaning through the image data of the camera set, which is of great significance to improving the automation and intelligence level of the security monitoring system and ensuring the stability and order of social production and life. The existing human behavior recognition method needs to upload the image data collected by the camera to the cloud server first, save a large amount of video data in the cloud server, and manually view the video to label the data. [0003] In related technologies, in order to reduce the workload of manual labeling, the technical route of self-supervised learning method is adopted. However, on the one hand, when there are situati...

AI Technical Summary

Problems solved by technology

However, on the one hand, when there are situations such as object occlusion or human body occlusion, the recognition accuracy of the self-supervised learning method is low

On the other hand, the self-supervised learning method runs on the end-cloud server, which occupies a large amount of computing resources of the cloud server, resulting in a high delay in the task of human behavior recognition.

Images