Driving Alertness System Using Deep Learning

Vaidya, Prof. R. S. · 2024 · Crossref

DOI: 10.22214/ijraset.2024.58722

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 addresses the critical safety issue of driver fatigue, a leading cause of road accidents characterized by compromised reaction times and impaired judgment. The authors aim to develop a robust Driving Alertness System (DAS) using deep learning to detect drowsiness in real-time. The research is motivated by the limitations of conventional warning systems and the need for precise, continuous monitoring of driver states. The study categorizes existing detection methods into physiological, vehicular, and behavioral approaches, highlighting the potential of computer vision to analyze facial expressions, specifically eye closure and yawning, as reliable indicators of fatigue. The proposed methodology integrates Mediapipe Face Mesh technology with a binary classification neural network. Mediapipe is utilized for the instant capture and identification of complex facial structures, allowing for the precise isolation of eyes and mouth landmarks despite dynamic driving conditions and varying facial features. This approach overcomes limitations associated with traditional methods that struggle with small eyes or inconsistent lighting. The system tracks subtle changes in eye opening/closing patterns and mouth movements to assess fatigue levels. To enhance model performance, the authors employ data standardization techniques, including mean averaging, scaling, and min-max scaling, to normalize input features. This preprocessing ensures that the binary classification neural network can effectively learn complex patterns distinguishing between alert and fatigued states, thereby improving convergence and generalization capabilities. The paper includes a literature survey comparing six recent studies on drowsiness detection, evaluating their advantages and disadvantages. For instance, it notes that hybrid machine learning approaches can achieve high accuracy (83%) but suffer from high time complexities, while other deep learning methods may overlook non-ocular facial features. The proposed system aims to mitigate these issues by focusing on comprehensive facial landmark tracking. The authors emphasize that the system’s ability to adapt to individual facial differences and monitor multiple behavioral cues simultaneously provides a more reliable detection mechanism than single-feature approaches. The significance of this work lies in its contribution to enhancing road safety through automated, real-time fatigue detection. By leveraging deep learning and advanced computer vision, the system offers a scalable solution for identifying drowsy drivers before accidents occur. The authors conclude that future research should focus on personalizing DAS algorithms to account for individual driving behaviors and integrating multiple biometric technologies for a holistic assessment of driver condition. This approach promises to create more accurate, user-friendly, and integrated safety systems, ultimately reducing accidents caused by driver impairment.

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
• distraction detection algorithms
• drowsiness
• gaze based attention detection
• dms validation
• truck driver fatigue

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