DREAMER: A Database for Emotion Recognition Through EEG and ECG Signals From Wireless Low-cost Off-the-Shelf Devices

Area of Science:

• Affective computing
• Biomedical signal processing
• Human-computer interaction

Background:

• Affect recognition is crucial for understanding human emotions.
• Existing methods often rely on expensive, non-portable medical equipment.
• There is a need for accessible, real-world affect recognition solutions.

Purpose of the Study:

• To introduce DREAMER, a multimodal database of EEG and ECG signals.
• To establish a baseline for affect recognition using low-cost, wearable sensors.
• To evaluate the feasibility of everyday affect recognition applications.

Main Methods:

• Recorded EEG and ECG signals from 23 participants during audio-visual stimuli.
• Collected self-assessments of valence, arousal, and dominance.
• Utilized supervised classification with Support Vector Machines (SVMs) on fused features.
• Validated self-assessments against a previous study.

Main Results:

• Achieved affect recognition performance comparable to studies using medical-grade devices.
• Demonstrated the effectiveness of portable, wireless, low-cost sensors.
• Established participant-wise affect recognition baselines for EEG and ECG fusion.

Conclusions:

• Low-cost, wearable devices are viable for affect recognition.
• The DREAMER database supports research into everyday affective computing.
• Publicly available data will facilitate further evaluation of sensor suitability.