Detection of Driver’s Drowsiness in a Video based on Deep Features

Nowadays, the existence of an intelligent system in high-end vehicles can reduce the vast number of fatal crashes around the world. Drowsiness accounts for a large portion of on-road vehicle crashes. Thus, the drowsiness of the driver has been considered a crucial factor in preventing an extension of fatal crashes. Drowsiness creates impairment in the real-time scenario, leading to fatalities. To reduce these fatal crashes, we propose a framework to detect the drowsiness of a driver by considering deep features of the driver’s facial expressions. In this framework, we used the transfer learning technique to extract deep features from each frame, and best among4 Classifiers are used for classification of a drowsy state i.e. Logistic Regression (LR), Support Vector Machine (SVM), Decision Tree (DT) and Random Forest (RF)are used for classification of a drowsy state.The experiment is performed on the National Tsing Hua University Driver Drowsiness Detection (NTHUDDD) benchmark dataset, and the obtained results are efficient compared with state-of-the-art techniques.