A Computer Vision Based Approach for Real-Time Drowsiness Detection Using Eye Aspect Ratio Analysis

Drowsiness can cause significant problems in conducting tasks that require ultimate attention like driving or operating heavy machines. It is important to detect the symptoms timely in order to ensure safety as it might lead to accident. The study revolves around introduction of a new method to identify drowsiness by analysing the eye aspect ratio (EAR) using computer vision technique. EAR measures eye openness by calculating distance between facial landmarks like the corners of the eyes and nose. As drowsiness sets in, eye lids sloop, causing EAR to drop this system uses pre trained deep learning models for real time landmark detection and ear calculation. EAR changes over time are tracked with a rolling window technique to detect drowsiness. Built with Python, open CV and Dlib, the system process accurately and quickly with minimal false alarms. It has been tested on various data sets and is effective in applications like transport, healthcare where alertness is of prime importance. Future improvement will involve adapting to different environment and integrating additional sensors for better accuracy. This approach helps preventing accidents caused by drowsiness.