Predicting individual emotion from perception-based non-contact sensor big data

Area of Science:

• Human-Computer Interaction
• Affective Computing
• Environmental Sensing

Background:

• Understanding human emotions is crucial for optimizing work environments.
• Existing methods for emotion detection often rely on direct user input or complex physiological monitoring.
• There is a need for unobtrusive methods to assess emotional states in real-time.

Purpose of the Study:

• To propose and validate a novel system for estimating individual emotions using indoor environmental data.
• To investigate the efficacy of wireless sensor data in predicting human emotional states.
• To establish a foundation for emotion-aware intelligent environments.

Main Methods:

• Development of wireless sensor nodes to collect indoor environmental parameters (e.g., temperature, humidity).
• Collection of physiological data (pulse, skin temperature) as indicators of emotion.
• Integration of environmental and physiological data into a big data system.
• Application of machine learning algorithms to estimate emotions from collected data over 60 days.

Main Results:

• The proposed system achieved over 80% accuracy in emotion estimation.
• Utilizing multiple sensor types significantly enhanced prediction accuracy.
• Environmental data proved effective in determining emotional states.
• Demonstrated the feasibility of unobtrusive emotion detection.

Conclusions:

• Indoor environmental data, when combined with physiological signals, can accurately predict human emotions.
• The developed system offers a promising approach for affective computing in real-world settings.
• This research highlights the potential of environmental sensing for advancing human-computer interaction and well-being.