Real time video processing for detecting unexpected behaviours

Unexpected behaviours are human actions that are considered out of the norm. In this paper, the unexpected behaviours that will be focused on is fall. Fall is a common problem especially for elderly. A fall can result in serious injury, permanent disability or even death. However, the consequences o...

Full description

Saved in:
Bibliographic Details
Main Author: Yong, Chun Yee
Other Authors: Kai-Kuang Ma
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2020
Subjects:
Online Access:https://hdl.handle.net/10356/139310
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:Unexpected behaviours are human actions that are considered out of the norm. In this paper, the unexpected behaviours that will be focused on is fall. Fall is a common problem especially for elderly. A fall can result in serious injury, permanent disability or even death. However, the consequences of fall can be largely mitigated if the fall is detected soon enough. Hence, fall detection devices are developed to detect fall and alert relevant people of the incident. Fall detection devices are categorized into three types; wearable-based, ambient-based and vision-based fall detector. The proposed system is developed using vision-based approach. Body shape change and bounding box techniques are utilized to develop the fall detection algorithm on Python and OpenCV. Telegram Bot Platform is integrated into the algorithm as part of the alert system for the device. Fall simulations conducted on the algorithm yield an accuracy of 83% in detecting falls.