Heatmap-Based Method for Estimating Drivers' Cognitive Distraction

Musabini, Antonyo; Chetitah, Mounsif · 2020 · arXiv

archive: archived pipeline: cataloged verified

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

Summary

This study addresses the detection of cognitive distraction, where drivers look at the road but are mentally disengaged, a gap in current systems that primarily detect visual or manual distraction. The authors propose a novel image-based method using eye-gaze heatmaps to estimate cognitive load. Data were collected from five male drivers on a highway in France. Participants completed two laps: a neutral baseline and a distracted lap involving secondary tasks designed to increase mental workload, such as n-back games. A near-infrared camera tracked eye-gaze vectors, which were projected onto an imaginary "virtual wall" to generate heatmaps representing gaze dispersion. Features including pixel intensity histograms, geometric contours, and looking-ahead confidence were extracted. Support Vector Machine classifiers were trained using leave-one-driver-out cross-validation. Results showed distinct visual differences in gaze patterns: neutral driving produced wide-area heatmaps, while distracted driving resulted in narrowed, concentrated shapes. Classification accuracy improved with longer observation windows, reaching 85.2% at 60 seconds. A 30-second window achieved 81.4% accuracy. The method successfully discriminated between neutral and distracted states using only gaze data. The findings demonstrate that gaze dispersion heatmaps provide a viable, contactless approach for detecting cognitive distraction. This method offers discriminative power comparable to more invasive techniques, suggesting potential for integration into vehicle safety systems to monitor driver inattention.

Key finding

Heatmap-based representation of eye-gaze dispersion has discriminative power to recognize cognitive distraction: drivers explore a wider visual area in neutral driving compared to distracted driving, with SVM achieving 85.2% accuracy.

Methodology

lab_experiment

Sample size: 10

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. Discovered via discover_arxiv on 2026-05-04 (4 acquisition events logged).

StageOutcomeToolModelPromptAttemptsCompleted
discover success arxiv 3 2026-05-04
archive success 1 2026-05-04
extract success cached 2 2026-06-07
clean success 1 2026-06-01
chunk success 1 2026-06-01
embed success 1 2026-06-02
enrich success 1 2026-05-04
promote success 1 2026-05-04
summarize success llm qwen3.6-27b-prismaquant summ-v5 2 2026-06-07
tag success vector_similarity 17 2026-06-11
verify success 1 2026-05-08

Summary generated by qwen3.6-27b-prismaquant on 2026-06-07; verification: verified.

Topics

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

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).