AIoT-2023-Team-05-Kontogiannis-Velgakis

End-to-end Artificial Intelligence of Things project about developing a system for recognizing automatically six common phrases in American Sign Language (ASL). Using MetaMotionR, a wristband equipped with two sensors, a gyroscope and an accelerometer, kinesiological data of the gestures were recorded, processed and analyzed in order to train and evaluate three statistical classifiers and two neural network classifiers. The SVM-optimized classifier emerged as the most accurate, followed closely by the RandomForest and CNN classifiers.