Skip to content

Automated computer-vision analysis of gaze and proxemics from classroom video

Research Method · All research methods
Evidence · 1 claim (1 for) · 1 study (1 associational), q3 · 1 of 1 report an effect size · 1 claim rests on one study

Description

The study adapted the Sharingan gaze-estimation framework (Tafasca et al., 2024), a pretrained transformer-based model reported with AUC 0.944, to derive joint attention and mutual gaze frame by frame from standard 2D video. A separate pipeline combining YOLOv11 pose estimation, DeepOCSort tracking, and depth-based distance computation computed pairwise interpersonal distances, categorized via 33rd/66th percentile thresholds into close, neutral, and far levels. The authors present this as a nonintrusive, scalable methodological alternative to intrusive sensors, lab-only setups, and manual coding in authentic classrooms.

Accounts

  • Automated computer-vision pipeline for extracting gaze and proxemics from ordinary 2D classroom video: The study adapted the Sharingan gaze-estimation framework (Tafasca et al., 2024), a pretrained transformer-based model reported with AUC 0.944, to derive joint attention and mutual gaze frame by frame from standard 2D video. A separate pipeline combining YOLOv11 pose estimation, DeepOCSort tracking, and depth-based distance computation computed pairwise interpersonal distances, categorized via 33rd/66th percentile thresholds into close, neutral, and far levels. The authors present this as a nonintrusive, scalable methodological alternative to intrusive sensors, lab-only setups, and manual coding in authentic classrooms. (Ridwan Whitehead · Andy Nguyen · Sanna Järvelä (2026))

Claims

Key Sources

  • Ridwan Whitehead · Andy Nguyen · Sanna Järvelä. (2026). How gaze and proxemics shape and signal shared regulatory processes during AI-supported collaborative learning. Intern. J. Comput.-Support. Collab. Learn. https://doi.org/10.1007/s11412-026-09478-7
Page metadata
Field Value
type research-method
id automated-computer-vision-analysis-of-gaze-and-proxemics-from-classroom-video
title Automated computer-vision analysis of gaze and proxemics from classroom video
description The study adapted the Sharingan gaze-estimation framework (Tafasca et al., 2024), a pretrained transformer-based model reported with AUC 0.944, to derive joint attention and mutual gaze frame by frame from standard 2D video.
status review
generated by process:settle-candidates on 2026-10-18
trust unverified — no one has confirmed this page's content

Frontmatter is the page's machine-readable record — see the Schema & Guide.