IMU-based Spectrogram Approach with Deep Convolutional Neural Networks for Gait Classification

We propose a wearable sensor-based gait classification system. Our approach assumes that multiple IMU sensors attached to various body parts can capture the gait characteristics that are used to predict whether the subject has foot abnormalities or athletic performance. We first transform raw sensor signals into spectrogram images and feed this visual representation to deep CNN models. The IMU data were acquired from 7 sensors attached to the pelvis, thighs, shanks, and feet while 69 people in three groups walked on a 20-m straight path. We investigated classification accuracy according to the number and location of attached sensors to optimize performance. Our experimental results show that only a single IMU sensor data can successfully predict the subject groups even without requiring hand-craft extraction and selection of features. As a result, practical applications could be easily deployed with less energy consumption. We plan to generalize our approach to predicting various health status information of people living in the ambient intelligent environment.