Two-stage facial-expression recognition pipeline for synchronous online-learning video

Research Method · All research methods
Evidence · 1 claim (1 for) · 1 study (1 associational), q2 · 0 of 1 report an effect size · 1 claim rests on one study

Description

The article's analytical pipeline processes session recordings through a two-shot detector: img2pose pre-trained on the Wider face dataset performs facial localization, and Residual Masking Network classifies emotion into seven classes (Anger, Disgust, Fear, Happiness, Sadness, Surprise, Neutral). The authors chose a two-step approach because a recent survey suggested it "achieves better performance on out-of-sample data", and relied on pre-trained models because small collaborative-learning datasets (< 100 examples per class) preclude training domain-specific models. Residual Masking Network achieved 73.28% accuracy on FER2013, above reported human recognition accuracy of 65.5%.

Accounts

• FER analysis pipeline combining img2pose facial localization and Residual Masking Network emotion detection for synchronous online learning video: The article's analytical pipeline processes session recordings through a two-shot detector: img2pose pre-trained on the Wider face dataset performs facial localization, and Residual Masking Network classifies emotion into seven classes (Anger, Disgust, Fear, Happiness, Sadness, Surprise, Neutral). The authors chose a two-step approach because a recent survey suggested it "achieves better performance on out-of-sample data", and relied on pre-trained models because small collaborative-learning datasets (< 100 examples per class) preclude training domain-specific models. Residual Masking Network achieved 73.28% accuracy on FER2013, above reported human recognition accuracy of 65.5%. (Duong Ngo et al. (2024))

Claims

• Emotional synchrony frequency differs slightly across regulatory activities but the metric does not discriminate SSRL, cognitive, and task-execution interactions [+M]

Related Research Methods

-

Key Sources

• Duong Ngo, Andy Nguyen, Belle Dang, Ha Ngo. (2024). Facial Expression Recognition for Examining Emotional Regulation in Synchronous Online Collaborative Learning. International Journal of Artificial Intelligence in Education 34:650–669. https://doi.org/10.1007/s40593-023-00378-7

Page metadata

Field	Value
type	research-method
id	two-stage-facial-expression-recognition-pipeline-for-synchronous-online-learning-video
title	Two-stage facial-expression recognition pipeline for synchronous online-learning video
description	The article's analytical pipeline processes session recordings through a two-shot detector: img2pose pre-trained on the Wider face dataset performs facial localization, and Residual Masking Network classifies emotion into seven classes (Anger, Disgust, Fear, Happiness, Sadness, Surprise, Neutral).
status	review
generated	by process:settle-candidates on 2026-10-27
trust	unverified — no one has confirmed this page's content

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