Elderly Fall Detection and tracking

Abstract

The title ”Elderly Fall Detection and tracking” suggests that the project involves the application of deep learning techniques for the purpose of tracking Elder persons fall detection and identifying the fall as well as live detetction and tracking.In my work it is primarily focused on the Elderly people. Both people identification and tracking face challenges such as occlusion, scale variations, and real-time processing requirements. Deep learning techniques have shown significant advancements in addressing these chal- lenges. There are number of algorithms and pretrained models that are used for tracking of people in relatime video like mediapipe, openpose, posenet, YOLO. Through all of these models we can detect persons by calculating the keypoints of the human body. Still there are number of preprocessing required for performing traking in realtime. In my project work I have implemented the model mediapipe for human body detection and tracking in closed room. for the processing of the image calculation of the human body keypoints and generation of the csv file has been done. Experimental results shows that for human identification and tracking mediapipe works better than any other models that are used for people identification.