/Monitoring-Driver-Drowsiness-from-physiological-and-visual-aspect-using-

A drowsiness detection system using OpenCV and CNN Methodology to provide safety and accident prevention.

Primary LanguageJupyter NotebookMIT LicenseMIT

Monitoring-Driver-Drowsiness-from-physiological-and-visual-aspect-using- CNN and OpenCV

Abstract The primary objective is to develop a system that can reliably detect drowsiness in a driver based on eyelid movement and yawning and then provide the driver with accurate audio notifications in real time. Another objective is to develop a method for automatically monitoring the driver's ocular retina in order to detect signs of fatigue. The driver should be alerted by the system if they yawn excessively or shut their eyes for a few seconds. We conclude that one may properly assess a driver's tiredness level by building a hybrid drowsiness detection system that combines non-intrusive physiological indicators with other metrics. Many accidents on the road may be prevented if drivers were notified when they began to show signs of sleepiness.

Keywords: Driver Fatigue, Driver Distraction, Sensors, Drowsiness Detection, Facial Expressions, Behavioural Measures, Machine Learning, Deep Learning, Spatiotemporal Features, OpenCV, CNN

Datasets Link

  1. https://www.kaggle.com/datasets/dheerajperumandla/drowsiness-dataset
  2. https://www.kaggle.com/datasets/rakibuleceruet/drowsiness-prediction-dataset
  3. https://www.kaggle.com/datasets/adinishad/prediction-images
  4. https://www.kaggle.com/datasets/serenaraju/yawn-eye-dataset-new
  5. http://mrl.cs.vsb.cz/eyedataset

Results & Conclusions -

S. No. Model Accuracy

1 Baseline Model 73.20%

2 Final Model 77.84%

3 Gray Wolf Opt. 91.2%