Sensor-Based Hand Gesture Detection and Recognition by Key Intervals

This study aims to present a novel neural network architecture for sensor-based gesture detection and recognition. The algorithm is able to detect and classify accurately a sequence of hand gestures from the sensory data produced by accelerometers and gyroscopes. Each hand gesture in the sequence is regarded as an object with a pair of key intervals. The detection and classification of each gesture are equivalent to the identification and matching of the corresponding key intervals. A simple automatic labelling is proposed for the identification of key intervals without manual inspection of sensory data. This could facilitate the collection and annotation of training data. To attain superior generalization and regularization, a multitask learning algorithm for the simultaneous training for gesture detection and classification is proposed. A prototype system based on smart phones for remote control of home appliances was implemented for the performance evaluation. Experimental results reveal that the proposed algorithm provides an effective alternative for applications where accurate detection and classification of hand gestures by simple networks are desired.