Developing a fast and efficient hand tracking technique could help determine whether the amount of hands on a wheel affects a drivers ability to drive. This hand monitoring can be paired with the drivers accident history to assess the assertion that less hands on a wheel correlates with a greater chance of accident occurrence. This analysis investigates the best way to achieve hand classification from a real-time video.
Specifically, this investigation will decide the best detector for hand detection, the optimal tracker for hand tracking, and a reliable method for hand classification.