Method Implementation Of Indoor Fall Detection System For The Elderly Based On Hybrid Mode

According to the seventh census in China,the current population over 60 years old in China accounts for 18.7%,the aging phenomenon is intensifying,it is expected that home care will be one of the main pension methods in the future,statistical research shows that falls are one of the biggest threats to the elderly living alone,how to ensure the home safety of the elderly living alone has become one of the main research hotspots.With the proposal of various advanced AI algorithms,it is of great research significance to quickly and accurately identify the abnormal fall behavior of the elderly in surveillance video by using a multi-algorithm fusion scheme to solve the fall problem of the elderly living alone.Based on Alpha Pose algorithm and emotion recognition technology,this paper proposes a video fall recognition scheme,which maximizes the use of audio and video information in the monitoring scene,through face recognition,fall judgment and emotion recognition,when the elderly living alone fall,the identity information of the elderly and whether they need to be rescued can be sent to the family and medical institutions in time,so that medical institutions can take corresponding rescue measures in time according to the identity of the faller,the main research content of this paper is as follows:(1)Aiming at the fact that the original fall detection algorithm is prone to misjudgment,this paper designs a decision-level fusion algorithm,which fuses the fall detection results obtained based on Alpha Pose model and the emotion recognition results obtained based on LSTM model,and obtains the weighting coefficients of different algorithm results through experiments to ensure the accuracy of the final fall judgment.Experiments show that compared with a single Alpha Pose model,the proposed method improves the decision accuracy.(2)Based on the fusion algorithm improved in this paper,a monitoring system scheme of audio and video joint fall judgment is designed and implemented,which collects surveillance video in indoor environment and realizes functions such as fall action,identity information and emotion recognition.The experimental results show that the system has high detection accuracy and can work in real time,which meets the expected design requirements.