Abstract:Emotions are central for human intelligence and should have a similar role in AI. When it comes to emotion recognition, however, analysis cues for robots were mostly limited to human facial expressions and speech. As an alternative important non-verbal communicative fashion, the body gesture is proved to be capable of conveying emotional information which should gain more attention. Inspired by recent researches on micro-expressions, in this paper, we try to explore a specific group of gestures which are spontaneously and unconsciously elicited by inner feelings. These gestures are different from common gestures for facilitating communications or to express feelings on ones own initiative and always ignored in our daily life. This kind of subtle body movements is known as ‘micro-gestures’ (MGs). Work of interpreting the human hidden emotions via these specific gestural behaviors in unconstrained situations, however, is limited. It is because of an unclear correspondence between body movements and emotional states which need multidisciplinary efforts from computer science, psychology, and statistic researchers. To fill the gap, we built a novel Spontaneous Micro-Gesture (SMG) dataset containing 3,692 manually labeled gesture clips. The data collection from 40 participants was conducted through a story-telling game with two emotional state settings. In this paper, we explored the emotional gestures with a sign-based measurement. To verify the latent relationship between emotional states and MGs, we proposed a framework that encodes the objective gestures to a Bayesian network to infer the subjective emotional states. Our experimental results revealed that, most of the participants would do ‘micro-gestures’ spontaneously to relieve their mental strains. We also carried out a human test on ordinary and trained people for comparison. The performance of both our framework and human beings was evaluated on 142 testing instances (71 for each emotional state) by subject-independent testing. To authors’ best knowledge, this is the first presented MG dataset. Results showed that the proposed MG recognition method achieved promising performance. We also showed that MGs could be helpful cues for the recognition of hidden emotional states.

Werewolf-XL: A Database for Identifying Spontaneous Affect in Large Competitive Group Interactions

Building a Chinese Natural Emotional Audio-Visual Database

SEWA DB: A Rich Database for Audio-Visual Emotion and Sentiment Research in the Wild

HEU Emotion: a Large-Scale Database for Multimodal Emotion Recognition in the Wild

A Database for Emotional Interactions of the Elderly.

Analyze Spontaneous Gestures for Emotional Stress State Recognition: A Micro-gesture Dataset and Analysis with Deep Learning

BNU-LSVED 2.0: Spontaneous Multimodal Student Affect Database with Multi-Dimensional Labels

HEU Emotion: A Large-scale Database for Multi-modal Emotion Recognition in the Wild

Affective Behaviour Analysis via Integrating Multi-Modal Knowledge

Aff-Wild Database and AffWildNet

An Effective Ensemble Learning Framework for Affective Behaviour Analysis

Analyses of a Multimodal Spontaneous Facial Expression Database

MAFW: A Large-scale, Multi-modal, Compound Affective Database for Dynamic Facial Expression Recognition in the Wild

Development of a Chinese werewolf deception database

"Biovid Emo DB": A Multimodal Database for Emotion Analyses Validated by Subjective Ratings.

MEAD: A Large-Scale Audio-Visual Dataset for Emotional Talking-Face Generation

Casme Ii: an Improved Spontaneous Micro-Expression Database and the Baseline Evaluation

The design and evaluation of multimode affective database

Multimodal Spontaneous Emotion Corpus for Human Behavior Analysis

A Chinese Multi-label Affective Computing Dataset Based on Social Media Network Users

An EEG-Based Multi-Modal Emotion Database with Both Posed and Authentic Facial Actions for Emotion Analysis