Driver activity recognition through hand gestures master thesis

Driver activity recognition and monitoring is a widely researched topic. One of the primary reasons for vehicle accidents is distracted drivers. The most researched observation method are eye and iris detection, where the prediction is based on the driver’s gaze. Understanding what drivers are doing while driving could help develop systems to prevent further accidents. Throughout this article, we explore hand gestures and how these can be used to determine the activity of the driver, due to the natural occurring interaction inside the car include distinct hand gestures, e.g. holding the wheel or the gear stick. With an average accuracy of 85.60% based on 17 evaluations in two cars with 13 participants and six different gestures, we conclude that hand gestures provide enough insight to accurately classify driver activities. ACM Classification