Driver Drowsiness Detection Using Eye Movement Behavior

Biradar, Sonali; Pathak, Shruti; Shirude, Nikita; Tiwari, Priya; Patil, Uma · 2025 · Crossref

DOI: 10.36948/ijfmr.2025.v07i03.44117

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the critical safety issue of driver drowsiness, which contributes to approximately 20% of global fatal traffic accidents. The authors argue that conventional methods, such as subjective self-reports and scheduled breaks, are inadequate because they rely on a driver’s often impaired ability to recognize their own fatigue levels. To overcome this, the paper proposes a non-intrusive, real-time monitoring system that utilizes computer vision and machine learning to detect drowsiness through eye movement behavior. The primary objective is to create a predictive model that can alert drivers before they reach a state of critical fatigue, thereby reducing accident risks. The methodology centers on a system architecture that processes live video feeds from a webcam to monitor facial features. The system employs Haar Cascade classifiers, a robust object detection algorithm, to identify the driver’s face and eyes in real-time. Key metrics for drowsiness detection include the Eye Aspect Ratio (EAR), which measures eye openness, and the duration of eye closure. The system calculates a drowsiness score based on the percentage of time eyes remain closed; if this score exceeds a safety threshold, an audio-visual alert is triggered. The experimental validation utilized both webcam-recorded videos and public datasets, including the CEW (Closed Eyes in the Wild) and Yawn Dataset, testing the system under varying lighting conditions and face angles. The results demonstrate that the proposed system effectively detects early signs of fatigue, such as microsleeps and yawning. Under normal lighting conditions without facial obstructions, the eye detection module achieved an accuracy of approximately 88%, while yawn detection reached 84% accuracy. The system maintained real-time performance at 20–24 frames per second on standard hardware (Intel i5 CPU, 8GB RAM) without requiring GPU acceleration. However, the study identified limitations, including reduced accuracy in poor lighting, shadows, or when faces were partially occluded by glasses or hands. Additionally, yawning detection occasionally produced false positives when drivers were speaking or laughing. The significance of this work lies in its demonstration that lightweight, efficient image processing techniques can provide a practical foundation for real-time drowsiness detection in vehicles. The authors conclude that while Haar cascades offer a viable low-cost solution, future improvements should involve advanced deep-learning models, such as Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, to enhance robustness under challenging conditions. The study also suggests that integrating multiple indicators, including head position and blinking patterns, will further improve prediction accuracy. This research contributes to the field of vehicular safety by providing a scalable, non-intrusive framework for monitoring driver alertness.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• drowsiness detection algorithms
• drowsiness
• microsleep
• gaze based attention detection
• dms validation
• distraction detection algorithms

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: physiological data
• Methodological Resource: tool software, measurement protocol