Design and Development of a Non-Contact ECG-Based Human Emotion Recognition System Using SVM and RF Classifiers

Area of Science:

• Human-Computer Interaction
• Biomedical Engineering
• Machine Learning

Background:

• Emotion recognition is crucial for advancing human-machine interaction (HMI).
• Emotionally intelligent systems can enhance daily routines and aid in mental state assessment.
• Early detection of emotional states may help prevent mental health issues like depression.

Purpose of the Study:

• To propose a unimodal emotion classifier using electrocardiogram (ECG) signals.
• To evaluate the performance of machine learning models in classifying human emotions from ECG data.
• To assess the feasibility of using non-contact ECG for emotion recognition.

Main Methods:

• Acquired non-contact ECG signals using a capacitive sensor-based ECG belt.
• Developed and trained Support Vector Machine (SVM) and Random Forest (RF) classifiers.
• Utilized 10-fold cross-validation on ECG data from 45 subjects across three age groups.

Main Results:

• The Random Forest (RF) classifier achieved a minimum accuracy of 98.2%.
• The Support Vector Machine (SVM) classifier achieved a minimum accuracy of 86.6%.
• Both models demonstrated effective emotion classification capabilities from ECG signals.

Conclusions:

• Machine learning models can effectively classify human emotions using non-contact ECG signals.
• The proposed system shows promise for developing advanced emotion-aware HMI.
• High classification accuracies indicate the potential for real-world applications in mental state monitoring.