Driver Drowsiness Detection Using Visual Behavior and ML

Mr .S.RAMA KRISHNA1, MALISETTY KALYANA RAGHAVA 2, MUDDPATI VISHNU SAI HARSHA VARDHAN3, NAKKA UDAY KIRAN4,PENUMATCHA RAMA KRISHNA BAPI RAVI TEJA VARMA 5

Assistant Professor of CSE-Data Science, KKR & KSR Institute of Technology and Sciences. 1

BTech CSE-Data Science, KKR & KSR Institute of Technology and Sciences, Guntur, Andhra Pradesh, India.2-5

ABSTRACT
Driver drowsiness is an important reason for most accidents that lead to serious injuries and deaths. This project is aimed at a driver drowsiness detection system that uses machine learning for determination of visual behavioral patterns. The system takes video using a camera in real time, runs OpenCV for processing, and uses their Haar Cascade classifier for detection of landmarks in the face. Factors such as blinking, yawning, and head position put in place to enhancement detection of drowsiness states of the driver. Being modeled, with high accuracy of detection and classification of drowsiness states is attended by the use of Convolutional Neural Network(CNN).

This solution envisages an extensive range of adaptability in real-world applications, including but not limited to long- haul transportation and fleet management. The system may be well-suited for embedding in automated vehicle safety systems or in connection with existing advanced driver- assistance systems (ADAS). Thus, it shall act continuously in monitoring driver behavior, producing real-time alerts with the goal of eliminating cases of drowsy driving and therefore improving general road safety.

In the end, this project articulates an effective, real-time driver drowsiness detection, which is based on leading computer vision and deep learning techniques. Different technologies invoked here should make the system attain better efficiency and accuracy while being scalable for detecting issues due to fatigue. Improvements may be realized in the future by infusion into the detection models for edge AI and be used for real-time processing on embedded devices in order to make them more general for usability in vehicle applications.

Keywords:Html,css,opencv,pandas,,numpy,CNN.keras,m achine learning,real-time detection