SMARTEYE - for human posture and activity monitoring Part 2

• Main Author: Lim, Jay Boon Kiat
• Other Authors: Deepu Rajan
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/69116
• Institution: Nanyang Technological University

The purpose of this study is to design a system that allows monitoring of the elderly within their household at all times. The report consists of three stages for monitoring – Human Detection, Human Posture Recognition, and Human Activity Recognition. Human detection allows the system to detect the exact location of the elderly whereas, posture recognition and activity recognition will determine the current status of the elderly. Two postures – Standing and Lying, are recognized by the system. By using the posture recognition algorithms, the system will classify the ongoing activity based on the sequence of postures detected. The system will be implemented with background subtraction and Histogram of Oriented Gradients. Information retrieved from the subtracted background and the classified posture will be used to classify the activity. After testing the implemented system, a detection accuracy of 90% has been achieved for human detection and an accuracy of 93% has been achieved for standing posture detection. Although, an accuracy of only 28% has been achieved for lying posture detection, the implemented system managed to detect falling activities accurately. Therefore, a conclusion can be drawn such that the current status of the elderly can be monitored successfully using the implemented system.