Drowsiness Detection Using CNN-LSTM

Driver fatigue is a major reason to road accidents, prompting the need for systems which can detect drowsiness in real time and prevent potential mishaps. It presents a deep learning-based solution utilizing CNN and LSTM networks to accurately find drowsiness based on the video inputs of a driver’s face. By analyzing eye and facial expressions, it identifies early signs of fatigue then triggers an alert. The model, trained on a labeled dataset which contains drowsy and non-drowsy drivers, achieves an accuracy of 97.83%. Real-time video processing is powered by OpenCV and advanced facial recognition techniques. The system is implemented on a Raspberry Pi 3, which runs the detection code and displays an alert on an external display board mounted outside the vehicle. When drowsiness is detected, an alert symbol is shown to notify nearby vehicles and reduce potential accidents. This solution demonstrates the potential of combining CNNs for feature extraction part and LSTMs work is to find temporal pattern analysis to enhance road safety and mitigate fatigue related accidents.