Fall detection system design based on visual recognition and multi-sensor
Aiming at the problem that accidentally fall of elderly can not be rescued in time,a fall detection system based on visual recognition and multi-sensor is designed.Taking Raspberry Pi as the core,the proposed system,cameras and sensors,positioning module,heart rate module,and narrow band Internet of Things(NB-IoT) module are used to collect and transmit information and the elderly fall detection indoor and outdoor are accomplished.Threshold detection and SVM classification method are used to improve fall detection accuracy.Monitoring App is used to realize the elderly status and location information inquiry function.The experimental results demonstrate that the accuracy of fall detection indoors and outdoors is 92.8％ and 91.0％,respectively.The system can effectively detect the elderly fall behavior and send alarm and positioning information to their supervisors,which provide wise care for the elderly.
fall detectionsupport vector machinethreshold detectionOpenPose algorithm
万方数据
维普
