DROWSINESS DETECTION AND MONITORING SYSTEM

Ghole, Uzair; Chavan, Pravin; Gandhi, Siddharth; Gawde, Rohit; Fakir, Kausar · 2020 · Crossref

DOI: 10.1051/itmconf/20203203045

archive: archived pipeline: cataloged verified

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

Summary

This paper presents the design and implementation of a non-intrusive drowsiness detection and monitoring system for four-wheeler vehicles, motivated by the significant safety risks and fatalities associated with driver fatigue. While electroencephalogram (EEG) signals offer high accuracy for detecting wakefulness, they require intrusive headsets that are impractical for widespread adoption. To address this, the authors propose a computer vision-based solution that utilizes image processing to monitor the driver’s facial features, specifically eye status, in real-time. The system aims to detect signs of drowsiness and generate immediate alerts to prevent accidents, leveraging Internet of Things (IoT) and cloud computing technologies for data analytics and remote notification. The prototype hardware consists of a Raspberry Pi 3, a USB web camera, an alarm source, and necessary power and wiring components. The software stack runs on Raspbian OS and utilizes OpenCV for computer vision tasks. The methodology employs Haar Cascade classifiers for rapid face detection, followed by the identification of 68-point facial landmarks to locate specific features such as the eyes. The core metric for determining drowsiness is the Eye Aspect Ratio (EAR), which is calculated continuously from the video feed. To mitigate false positives caused by micro-napping or transient eye closures, the system implements a threshold mechanism based on the duration of the closed-eye state. Data collected from the system is transmitted to a cloud platform, specifically AWS Greengrass, for storage, analysis, and the generation of meaningful analytics. Experimental results demonstrate the system's ability to distinguish between wakeful and drowsy states using EAR values. Statistical analysis established an average EAR of 0.3 for a wakeful state. The system was configured with a threshold where an EAR below 0.3 for eight consecutive frames triggers a drowsiness alert. Testing data showed that EAR values of 0.348 and 0.291 corresponded to a "Wake" state, while values of 0.218 and 0.210, sustained for eight and ten frames respectively, were correctly identified as "Drowsy." The system generates two types of alerts: an in-car alarm to rouse the driver and a notification sent to emergency contacts via the cloud platform. A one-month study provided percentage distribution data, highlighting the potential for long-term monitoring and data-driven insights. The significance of this work lies in its practical, low-cost approach to enhancing road safety without requiring drivers to wear additional equipment. By combining real-time computer vision with IoT infrastructure, the system not only provides immediate feedback to the driver but also enables remote monitoring and historical data analysis. The authors conclude that proper implementation of such systems can significantly reduce accidents caused by driver fatigue, thereby saving lives. The study validates the feasibility of using EAR and frame-count thresholds as effective indicators for drowsiness detection in real-world driving scenarios.

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
• dms validation
• drowsiness
• microsleep
• 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