A Real-Time Position Monitoring System For Fall Detection and Analysis Using Human Pose Estimation

It has been observed that there was an increase in the afflictions and catastrophes faced by elderly individuals. Concussion and hemorrhage are often a result of a fall and this could prove to be fatal. The normal changes of aging like vision impairment, hearing difficulties, illnesses and physical conditions affect the strength and balance of elderly individuals. Difficult lighting conditions or throw rugs at home can cause elders to trip or slip resulting in a fall. In order to provide the victim of the fall with immediate aid, it is crucial to develop a system to detect falls in real-time. The system proposed in the paper aims to use various object detection algorithms, apply human pose estimation and use various fall detection algorithms on real-time surveillance footage to detect the occurrence of a fall. In addition to the above point, the paper's system proposes to detect the cause of the fall with the help of a wearable device. In case of an emergency, the next of kin gets notified about the occurrence of the fall along with the pulse and BP (Blood Pressure) rate. This helps in not only detecting if a fall has occurred but also helps to determine the cause of the fall.