Real-time fall detection for elderly with obstructions consideration using kinect

Amir Husin, Amir Hazim (2025) Real-time fall detection for elderly with obstructions consideration using kinect. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (Full Text) Real-time fall detection for elderly with obstructions consideration using kinect.pdf - Submitted Version Download (4MB)

Abstract

The increased number of elderly people is growing in tandem with improvements in healthcare. However, the risks that go along with it have also increased, including the number of falls. According to several studies, elderly persons fall at least once a year. For elderly persons aged 65 and older, falls are the leading cause of unintentional death. Many elderly people in Malaysia spend the day alone at home because their family members are either at work or school. Therefore, to properly identify the occurrence of falls, a fall detection system is required. The construction of such a system to identify when an individual falls or loses their balance is described in this thesis. To evaluate and spot patterns that point to a fall incident, the system uses Kinect sensors and algorithms to track a person's movements and postures. Skeletons and joints are identified and retrieved, including the heads, shoulders, hips, and left and right ankles. The Y-coordinate values and threshold values are obtained by implementing the fall algorithm. To ascertain fall status, the absolute values of the Y-coordinate and joints are compared to the threshold value. The system also differentiates between falls, sitting on a chair, sitting on the floor, and also fall incident behind an obstruction. When a possible fall incident is detected, the system activates an alarm. The expected results include high accuracy in detecting falls, minimizing false positives, and ensuring that the system operates effectively even when there is partial body visibility. The system aims to provide timely alerts to caretakers or emergency responders, significantly reducing response times and potentially saving lives. Additionally, the solution is designed to be minimally intrusive, ensuring it does not disrupt the elderly's daily activities.

Actions (login required)

View Item