Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

CMED: A Child Micro-Expression Dataset

Nikin~Matharaarachchi, Muhammad~Fermi Pasha, Sonya~Coleman, Kah PengWong

TL;DR

This work addresses the absence of child micro-expression data by introducing CMED, the first in-the-wild video dataset of spontaneous child micro-expressions collected during teletherapy-like Zoom sessions from 74 subjects. It details a full dataset creation pipeline—data collection, automated preprocessing, careful labelling with METT-based training, and a three-stage verification process—plus a 6-class and a 3-class variant (including No-ME). Baseline evaluations using LBP-TOP, VGG-16, and DGCNN show deep learning methods outperform handcrafted features, with DGCNN benefiting from body-landmark graphs and VGG-16 offering strong apex-frame-based performance for the 3-class variant. CMED’s real-world collection setting and comprehensive labeling enable robust analysis of child ME differences vs. adults and hold promise for improving computer-assisted therapy, especially in remote or rural contexts; access to the dataset is available upon request under a Data Access Agreement.

Abstract

Micro-expressions are short bursts of emotion that are difficult to hide. Their detection in children is an important cue to assist psychotherapists in conducting better therapy. However, existing research on the detection of micro-expressions has focused on adults, whose expressions differ in their characteristics from those of children. The lack of research is a direct consequence of the lack of a child-based micro-expressions dataset as it is much more challenging to capture children's facial expressions due to the lack of predictability and controllability. This study compiles a dataset of spontaneous child micro-expression videos, the first of its kind, to the best of the authors knowledge. The dataset is captured in the wild using video conferencing software. This dataset enables us to then explore key features and differences between adult and child micro-expressions. This study also establishes a baseline for the automated spotting and recognition of micro-expressions in children using three approaches comprising of hand-created and learning-based approaches.

CMED: A Child Micro-Expression Dataset

TL;DR

This work addresses the absence of child micro-expression data by introducing CMED, the first in-the-wild video dataset of spontaneous child micro-expressions collected during teletherapy-like Zoom sessions from 74 subjects. It details a full dataset creation pipeline—data collection, automated preprocessing, careful labelling with METT-based training, and a three-stage verification process—plus a 6-class and a 3-class variant (including No-ME). Baseline evaluations using LBP-TOP, VGG-16, and DGCNN show deep learning methods outperform handcrafted features, with DGCNN benefiting from body-landmark graphs and VGG-16 offering strong apex-frame-based performance for the 3-class variant. CMED’s real-world collection setting and comprehensive labeling enable robust analysis of child ME differences vs. adults and hold promise for improving computer-assisted therapy, especially in remote or rural contexts; access to the dataset is available upon request under a Data Access Agreement.

Abstract

Micro-expressions are short bursts of emotion that are difficult to hide. Their detection in children is an important cue to assist psychotherapists in conducting better therapy. However, existing research on the detection of micro-expressions has focused on adults, whose expressions differ in their characteristics from those of children. The lack of research is a direct consequence of the lack of a child-based micro-expressions dataset as it is much more challenging to capture children's facial expressions due to the lack of predictability and controllability. This study compiles a dataset of spontaneous child micro-expression videos, the first of its kind, to the best of the authors knowledge. The dataset is captured in the wild using video conferencing software. This dataset enables us to then explore key features and differences between adult and child micro-expressions. This study also establishes a baseline for the automated spotting and recognition of micro-expressions in children using three approaches comprising of hand-created and learning-based approaches.

Paper Structure

This paper contains 13 sections, 8 equations, 8 figures, 6 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Existing Micro-Expression Datasets
  3. Dataset Creation Methodology
  4. Data Collection
  5. Data Pre-processing
  6. Labeller selection and training to standardize
  7. Labelling Methodology
  8. Dataset Details
  9. Feature Exploration
  10. Evaluation
  11. Micro-expression Spotting
  12. Micro-expression Recognition
  13. Conclusion

Figures (8)

  • Figure 1: Dataset creation process
  • Figure 2: sample frames of raw videos. (a) frame showing only children and facilitators (b) frame showing screen share
  • Figure 3: sample frames after automated preprocessing. (a) exclusion criteria - face is covered (b) exclusion criteria - black bars or parts of other segments present in frame (c) & (d) minor intrusions
  • Figure 4: Sample apex frames from the dataset (a) anger (b) disgust (c) fear (d) sadness (e) surprise (e) happiness
  • Figure 5: The Region of Interest (ROI) and the base landmarks used for normalization.
  • ...and 3 more figures