3D Convolutional Neural Network for Gait Recognition And Demographic Factor Prediction Based on Plantar Pressure Images

This paper proposed a new gait recognition method based on 3D convolutional neural network(3D-CNN). The plantar pressure image is a continuous time series, and the current method has not been fully utilized the spatiotemporal information. By normalization and image stacking, the original pressure data is encoded into spatiotemporal feature vectors to implement classification recognition and regression prediction. We tested our algorithm on the CAD WALK Healthy Controls dataset which has 55 subjects and each subject has 24 walking images. The experiment results turn out that our 3D-CNN network can get high classification accuracy and demographic factors such as height, weight, age and shoe size can also be predicted better as compared to many other methods.