• affective computing
• inertial data
• machine learning
• wearable sensors

The aim of this thesis was to design and develop an AI-based system to classify human emotions based on inertial data (triaxial accelerometer and gyroscope) recorded using a smartwatch. An Android application was developed to collect recording sessions of inertial data during daily activities. Each recording session was associated with an emotion among disgust, anger, fear, sadness, anxiety, boredom and happiness, also specifying a level among low, medium, and high. Data were preprocessed and various classifiers were then trained whose levels of accuracy were compared to each other. The developed system is unobtrusive and privacy-preserving as it can help determine and track a person's mood only using inertial data.