Distracted Driver Detection and Driver Rating System using Deep Learning

The prevalence of driver distraction is on the rise due to the widespread adoption and complexity of in-vehicle technologies and portable devices. This trend poses a significant risk to road safety, as distractions and inattentiveness are major contributing factors to accidents. Therefore, the goal in this paper is to propose a methodology which can detect whether a driver is driving the car safely or performing an activity that may lead to an accident. For this, the primary source of data of drivers captured by an in-car camera, encompassing their face, arms, and hands. This dataset, which is openly accessible on Kaggle, is widely utilized for similar types of work in this field. In this work, deep learning and machine learning techniques are used to classify from the images whether the driver is distracted or not. By using these techniques, we will classify our input images into different categories of distraction in which our driver may be indulging in. The proposed model yields 98.5 % of validation accuracy and 0.0610 validation loss.