Fatigue Detection Using Artificial Intelligence and Computer Vision Algorithms

Friday, 13 February 2015
Exhibit Hall (San Jose Convention Center)
Rohit Goyal, Exton, PA
This project explores fatigue detection using artificial intelligence and computer vision algorithms. The work details how signs of fatigue, namely yawning, posture, and blinking, can be identified through algorithmic techniques. The technologies used are the Microsoft Kinect sensor system, typically employed in gaming, and the Open Source Computer Vision (OpenCV) libraries. The Microsoft Kinect API is used to detect the yawning and posture of an individual. Through the API, coordinates for lips are extracted and used as inputs to determine the extent to which a person is yawning. The positions for a person’s head, shoulders, and neck are also accessed through this API and are used to determine his or her posture angle. To determine whether or not someone is blinking, OpenCV libraries read images and locate faces and eyes in the frames. This work can be used to produce more cost-effective systems than current market options. Additionally, this research is important to many areas, including transportation, medicine, and strenuous professions. These findings can be synthesized with other technologies to determine how fatigue-related symptoms may affect driving capabilities, health problems, and occupational functionality.