Behaviour Recognition in an Elevator Using Kinect Sensor
Phan, Van Hoang Giang (2021)
Phan, Van Hoang Giang
2021
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2021112521721
https://urn.fi/URN:NBN:fi:amk-2021112521721
Tiivistelmä
Increasing population and urbanization has led to widespread use of elevators in buildings. There is always a concern associated with the safe usage of elevators. Manual video monitoring is often slow and ineffective when responding to emergency situations inside the elevator cabin, which can lead to undesirable consequences.
The objective of this thesis project was to develop a behaviour recognition prototype based on the Kinect depth sensor that could automatically recognize different behaviours in an elevator. Recognition results were recorded in the database and would be analysed by professionals to prevent emergency situations and to further improve safety and riding experience in an elevator.
The prototype utilized human silhouettes for recognizing behaviours. Project work comprised of designing gesture database, developing behaviour recognition software, testing and research on camera placement. For gesture database design stage, videos of gestures were recorded and tagged. In the software development stage, the behaviour recognition functionality of the prototype was implemented. During the testing stage, underlying algorithm’s performance was evaluated. In the final stage, case studies of camera placement for standing and lying gestures were carried out.
The developed prototype was able to detect defined behaviours with high accuracy. With further advancements, it would provide complete intelligent monitoring capability for elevators and further leverage their safety standards.
The objective of this thesis project was to develop a behaviour recognition prototype based on the Kinect depth sensor that could automatically recognize different behaviours in an elevator. Recognition results were recorded in the database and would be analysed by professionals to prevent emergency situations and to further improve safety and riding experience in an elevator.
The prototype utilized human silhouettes for recognizing behaviours. Project work comprised of designing gesture database, developing behaviour recognition software, testing and research on camera placement. For gesture database design stage, videos of gestures were recorded and tagged. In the software development stage, the behaviour recognition functionality of the prototype was implemented. During the testing stage, underlying algorithm’s performance was evaluated. In the final stage, case studies of camera placement for standing and lying gestures were carried out.
The developed prototype was able to detect defined behaviours with high accuracy. With further advancements, it would provide complete intelligent monitoring capability for elevators and further leverage their safety standards.