Human Activity Recognition Using 2D Convolutional Neural Networks

Abstract

Activity recognition represents the task of classifying data derived from different sensor types into one of predefined activity classes. Convolutional neural networks (CNN) as one of the best deep learning methods has recently attracted much attention to the problem of activity recognition, Most convolutional neural networks used for recognition task are built using convolution and pooling layers followed by a few number of fully connected layers. In this paper, we propose a new architecture based on 2D convolutional neural network that consists solely of convolutional layers and find that with removing the pooling layers and instead adding strides to convolution layers, the computation time will decrease notably while the model performance will not change or in some cases will even improve. Also its performance will be compared with some other handcrafted feature based methods. The third point that will be discussed in this paper is the impact of applying fast fourier transform to inputs before training learning algorithm. It will be shown that this preprocessing will enhance the model performance. Experiments on benchmark dataset demonstrate the high performance of proposed 2D CNN model with no pooling layers, achieving an overall accuracy of 95.69% on the test set.