Driver Vigilance Monitoring System

Driver doziness is a significant contributor to road accidents encyclopedically, challenging robust and real- time discovery systems. This paper presents a comprehensive motorist Alert Monitoring System using deep literacy and computer vision ways to descry motorist fatigue by assaying eye countries. The system employs a Convolutional Neural Network (CNN) trained on grayscale eye images to classify eye countries as' Open' or' Closed'. MediaPipe's Face Mesh result is employed for precise facial corner discovery and birth of eye regions, which are also preprocessed with CLAHE and regularized before feeding into the trained model. A critical aspect of the system involves covering the duration of unrestricted eye countries; if a predefined threshold is exceeded, an audio alarm is touched off to warn the motorist, along with visual cues on the display. The advanced system demonstrates high delicacy in eye state bracket and effective real- time doziness discovery, validated through comprehensive training and evaluation criteria including perfection, recall, F1- score, ROC AUC, and confusion matrices. A stoner-friendly Tkinter- grounded visual interface facilitates system control and provides real- time status updates and access to discovery logs, offering a practical result for enhancing road safety