Automated System for Detecting Distracted Driver

There has always been a safety issue while driving and this issue is of utmost importance with the increasing amount of accidents due to car crashing or distracted driving. As per the CDC motor vehicle safety division, one in five car accidents occurs due to distracted driving. Unfortunately, this led to over 425,000 people getting injured and around 3,000 people being killed due to distracted driving. This paper aims to monitor the behaviour of the driver and look for any kind of distraction. The model classifies the distractions into ten classes like talking on phone using left/right hand, interacting with the passenger, using the dashboard, drinking and reaching behind. Different architectures of Convolutional Neural Networks (CNN) like AlexNet and Inception-v3 are used which give promising results. The input to the model is the set of dashboard images with the help of which the behaviour of the driver is supervised for any distractions.