Emotional Body Gesture Recognition Based on the Instantaneous Pressure-Sensing Array Garment

Body gestures convey emotion-specific information. It is thereby an effective way to strengthen emotion recognition by recognizing emotional body gestures. In this work, we engineered a pressure-sensing array garment (PSAG) for the recognition of emotional body gestures. Our study focused on seating body gestures related to four emotions: calmness, happiness, anger, and sadness. PSAG was used to collect emotional body gesture data from ten subjects. Six distinct machine learning classifiers were then trained based on instantaneous pressure distribution to recognize the considered body gestures. Through performance evaluation and comparison of recognition models, we found that some could achieve good results in both person-dependent training/testing and cross-subject validation phases. For instance, the linear discriminant analysis model achieved a classification accuracy of 86.2% and 71.5% in the two mentioned phases, respectively. The study’s findings demonstrated the potential and effectiveness of our proposed approach in recognizing emotional body gestures. These recognized gestures can provide additional information for multi-modal emotion recognition.