Emotion Tracking Using Wristband

Abstract Real-time emotion tracking is essential for supporting mental health, enhancing education and better customer service. In this paper, we have developed an “Emotion Tracking Wristband” which is equipped with sensors, including Heart rate sensor, Galvanic Skin Response (GSR) sensors, and Skin Temperature sensor, to extract human emotions effectively. Wristband captures and analyzes the physiological data associated with diverse emotional states such as Happy, Angry, Sad, Neutral, Fear, Surprise and Disgust. Our approach involved inducing emotions in participants through video stimuli within real-life settings while concurrently collecting physiological data. Subsequently, we extracted a wide array of features from these signals, encompassing time, frequency, and nonlinear characteristics. These physiological signals are subsequently processed through SVM, KNN, Decision Tree and Random Forest machine learning algorithms, which have been meticulously trained on an extensive dataset of emotional patterns. This training empowers the system to accurately classify emotions, such as happy, angry, sad, neutral, fear, surprise and disgust. Ultimately, the “Emotion Tracking Wristband” bridges the gap between technology and emotion, opening up new avenues for understanding and managing our intricate emotional landscapes. Our system achieved an impressive overall accuracy rate of 83% across 30 participants, underscoring its efficacy in recognizing human emotions. Keywords— Wearable wristband, heartbeat, blood pressure, skin temperature, electrodermal activity, and emotion recognition etc.