Description

Real-time Distracted Driver Posture Classification using OpenCV

The number of road accidents due to distracted driving is steadily increasing.According to the World Health Organization global status report on road safety, traffic accidents are the eighth leading cause of death in the world, and nearly one-fifth of the traffic accidents were cause by driver distractions. 

In this project, we present a new dataset for “distracted driver” posture estimation. In addition, we propose a novel system that achieves 95.98% driving posturel estimation classification accuracy. The system consists of a genetically-weighted ensemble of Convolutional Neural Networks (CNNs). We show that a weighted ensemble of classifiers using a genetic algorithm yields in better classification confidence. We also study the effect of different visual elements (i.e. hands and face) in distraction detection and classification by means of face and hand localizations. Finally, we present a thinned version of our ensemble that could achieve a 94.29% classification accuracy and operate in a realtime environment.