COGNITIVE DRIVER ACTION RECOGNITION

Abstract

Operating an automobile is a multifaceted endeavor, demanding unwavering focus and attention from the driver. Distracted driving, encompassing any behavior that diverts the driver's concentration away from the road, poses a grave threat to road safety. Alarming statistics reveal that approximately 1.35 million lives are tragically lost each year due to road traffic accidents, inflicting significant economic ramifications, with road traffic crashes costing most nations an estimated 3% of their gross domestic product. The primary objective of our project is to put in place an exhaustive process for identifying potentially dangerous driving behaviors and discerning appropriate driving practices in light of this disappointing reality. By utilizing a wide range of machine learning models, we aim to correctly classify the given photos into discrete groups that correlate to various types of driver distraction. Furthermore, our work goes beyond simple classification; it also aims to perform a comparison analysis of different Machine Learning Models in order to determine how well they perform and how accurate they are in the context of cognitive driver action detection. This all-encompassing strategy demonstrates our dedication to improving traffic safety and lowering the possibility of collisions and injuries to other drivers. Key Words: Transfer learning, Deep learning, Image classification, Distracted driving, TensorFlow.