Feasibility study of emotion mimicry analysis in human-machine interaction.

Area of Science:

• Neuroscience
• Psychology
• Human-Computer Interaction

Background:

• Growing popularity of health apps for well-being.
• Emotional intelligence is key for social interaction and neurosensory health.
• Need for tools to aid emotion identification and expression.

Purpose of the Study:

• Develop and test a closed-loop feedback system for emotion projection via facial expression mimicry.
• Investigate the system's feasibility for therapeutic applications, especially for individuals with autism spectrum disorder.
• Enhance emotional intelligence and social interaction skills.

Main Methods:

• A digital mirror system displaying animated avatars expressing basic emotions.
• Users mimic avatar expressions, with a custom emotion recognition model analyzing facial feedback.
• Feasibility tested on 8 participants with 6 basic emotions and a neutral expression.

Main Results:

• Participants recognized emotions with 85.40% accuracy.
• Participants mimicked expressions with 46.67% accuracy.
• Initial 'surprise' expression deficiency addressed with ~30% performance boost through model enhancement.

Conclusions:

• The system demonstrates potential for enhancing emotional skill development and well-being.
• The digital mirror approach shows promise for therapeutic interventions.
• Further validation through larger studies is required.