A human behavior recognition method and recognition system based on deep neural network

Abstract

The invention provides a human body behavior recognition method based on a deep neural network, comprising: obtaining the original depth data stream of the actor; extracting the skeleton joint point data of the human body through the original depth data stream of the actor; using the extracted human skeleton joints The three-dimensional coordinates corresponding to the point data are used to model the entire human body; feature extraction is performed by modeling the entire human body, and the feature data is sent to the restricted Boltzmann machine network for preprocessing, and the weights obtained are initialized to the BP neural network Network parameters, train a deep neural network model, and perform behavior recognition on the result of feature extraction; use multi-threaded parallel processing to overlap the extracted human skeleton joint point data with the actual human body, and perform the recognized behavior Real-time display; build abnormal behavior template library and alarm the detected abnormal behavior. The invention can detect the change of the human body behavior in real time, and give an alarm to the abnormal behavior of the human body (such as falling).

Description

technical field [0001] The invention relates to the field of video recognition, in particular to a human behavior recognition method and recognition system based on a deep neural network. Background technique [0002] As we all know, now is the era of aging of the human body. my country has the largest elderly population in the world. According to reports, the elderly population will reach 248 million by 2020. Coupled with the implementation of family planning, the only child is busy with work. Lives independently most of the time. Many abnormal behaviors cannot be detected in time, so that the tragedies of the elderly delaying treatment and losing their lives occur from time to time, seriously affecting the quality of life of the elderly. For example, falling behavior is the direct cause of injury death among the elderly over 65 years old in my country. If we can accompany the empty-nest elderly in their daily life, effectively identify their behavior, and detect abnormal ...

AI Technical Summary

Problems solved by technology

[0005] In view of this, the purpose of the embodiments of the present invention is to provide a human behavior recognition method and recognition system based on a deep neural network, which aims to solve the problems of limited types of human behavior recognition, low recognition accuracy, and slow recognition speed in the prior art. question

Images