Emotion Analytics for Advanced Driver Monitoring System

From the recent advances in Driver Monitoring Systems (DMS) from automotive domain, research on Human Computer Interaction (HCI) based on emotion analytics has gained good interest from the research circles. Distraction and drowsiness will be causing more percentage of traffic accidents, but with the use of advanced DMS technology, we can significantly reduce these distractions and can make the driving a safer activity. Our proposed solution/approach with disguised emotion detection with analytics is enabled by machine learning and image processing algorithms to ensure that the detection of drowsiness or distraction is very accurate. The proposed method will inform the HMI system to provide an alert to wake up the driver if he or she is in drowsy state or take the proactive/necessary actions with the help of active safety systems. Emotion analytics is a technique which is used to analyze the emotion of an individual. It is used to recognize the change in the emotion. Deep Learning is used for the implementation of computer vision techniques which is implemented with the help of Convolutional Neural Network (CNN). In recent times, CNN has been successfully applied in analyzing visual images for many automotive applications. CNN model can be applied to recognize the emotion. We have trained CNN model with different depth using grayscale images. Emotions can be classified into following six categories i.e. Happy, Sad, Surprise, Angry, Neutral and Fear. After recognition, emotions are continuously analyzed. We recorded the emotion in particular time frame like how many times a person is Happy, Sad, Surprised etc. Standard & Tata Elxsi’s proprietary database is used for training the Emotion Recognition System. Proposed system is tested in Raspberry pi board and results found satisfactory. This analysis will help us to monitor the activity of driver. In case of any abnormal behavior we can take corrective measure to control the situation.