Driver Safety System using Internet of Things

Driver drowsiness or drunk driving is a major contributing factor in many road accidents around the world. We're all also aware of cases of drunk driving. We predict an increase in distraction-related collisions as wireless communication, entertainment, and driver assistance systems proliferate in the car industry. concentrating on preventing such accidents and preventing the loss of other lives The strategies for detecting driver inattentiveness and distraction based on behavioral data and deep learning algorithms are surveyed in this research. We were able to extract some helpful algorithms and strategies from the research papers that we might apply in our future work to improve accuracy and navigate challenges. These include yawning, face detection, head movements, entity detection, and blinking. However, because accurate and robust algorithms are required, developing a dependable and effective sleepiness detection system is a challenging task. The identification of driver distraction and sleepiness has been studied using a variety of techniques in the past. To evaluate how successfully these algorithms recognise drowsiness in light of the recent advancement of deep learning, modifications to these algorithms are necessary. We propose an Ensemble Averaging technique using state-of-the-art CNN models like Inception-V3 and VGG-16 which yielded us an accuracy of 97.5% on the validation data. This serves as the backbone to our main IOT system which will conduct a real time driver drowsiness, distraction and alcohol detection in the driver