A Cloud-Based Model for Driver Drowsiness Detection and Prediction Based on Facial Expressions and Activities

This paper proposes an efficient approach to detecting and predicting drivers' drowsiness based on the cloud. This work focuses on the behavioral as well as facial expressions of the driver to detect drowsiness. This paper proposes an efficient approach to predicting drivers' drowsiness based on facial expressions and activities. Four different models with distinct features were experimented upon. Of these, two were VGG and the others were CNN and ResNet. VGG models were used to detect the movement of lips (yawning) and to detect facial behavior. A CNN model was used to capture the details of the eyes. ResNet detects the nodding of the driver. The proposed approach also exceeds the results set by the benchmark mode and provides high accuracy, an easy-to-use framework for embedded devices in real-time drowsiness detection. To train the proposed model, the authors have used the National Tsing Hua University (NTHU) Drivers Drowsiness data set. The overall accuracy of the proposed approach is 90.1%.