What question did this study set out to answer?

The aim is to enhance automatic micro-expression recognition by addressing the challenges of subtle facial muscle movements and small datasets.

March 21, 2026Cognitive ComputationOpen Access

Resizer-Integrated RepGFPN-Based Multi-Scale Feature Fusion Network with Perturbation Training for Micro-Expression Recognition

View Full Paper

Ask AI

Bookmark

Authors

This research demonstrates an advanced method for recognizing micro-expressions, suggesting improvements in data efficiency and accuracy.

The aim is to enhance automatic micro-expression recognition by addressing the challenges of subtle facial muscle movements and small datasets.
Developed an IncepResizer module to modify input image sizes for better processing.
Utilized a lightweight backbone network for extracting multi-scale features effectively.
Implemented multi-scale feature interaction and fusion through the RMFNet module.
Employed a multi-layer perceptron (MLP) classifier for final classification.
Applied a perturbation training strategy to mitigate issues related to small sample sizes.
Achieved a utility F1 score (UF1) of 0.8916.
Obtained a Unweighted Average Recall (UAR) of 0.8913.
Demonstrated effective feature extraction capabilities even with limited data.

ZHAO et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark