Gesture Recognition Using Inertial Sensors with 1D Convolutional Neural Network
Zhenyu He, Qianying Li, Zhenfeng He · 2023
Nowadays, smart wearable devices are widely used and Micro-Electro-Mechanical Systems technology (MEMS) sensors are usually embedded in them. In this paper, a novel gesture recognition framework based on the 3D accelerometer and 3D gyroscope of smart phone is prpposed. Currently, most of the effort is focused on manually extracting features from MEMS sensors. However, without relevant domain knowledge, it is difficult to extract valid features to classify specific tasks. Instead of manual feature extraction, we propose 1D Convolutional Neural Network model to automatically extract features and gesture recognition from acceleration and gyroscope data. The best average recognition of ten Arabic numerals was 89.63%. The experimental results show that gesture-based Human-computer Interaction (HCI) can be applied as a new type of HCI for consumer electronics and mobile devices.