Multimodal Emotion Recognition of Hand-Object Interaction
summary
A 5 cm sensorized cube called iCube records capacitive touch (no pressure) and an accelerometer while 11 people perform action sequences; 237 trials produce 17 hand-crafted features (examples: average touch density and touch variability). Twelve features differ by emotion: anger and excitement engage larger contact area and more rotations, sadness yields longer interactions. Classification using tactile+kinematics reaches 75% accuracy across four emotions, showing the method works above chance.
AI-written summary · about
related talks
17mEMPress: Practical Hand Gesture Classification with Wrist-Mounted EMG and Pressure Sensing
11mAmbient Intelligence and Emotionally Intelligent Devices
49mAffective Computing – Bringing Humans and Machines Closer Through Emotions
22mSeeing, Sensing and Recognizing Laban Movement Qualities
50mAffective Computing
21mEmotion Estimation for Social Robotics
20mAdvancing Hand Gesture Recognition with High Resolution Electrical Impedance Tomography
31mRobotic emotions