CNN-based and DTW features for human activity recognition on depth maps

Jacek Trelinski,Bogdan Kwolek

DOI: https://doi.org/10.1007/s00521-021-06097-1

2021-05-12

Neural Computing and Applications

Abstract:Abstract In this work, we present a new algorithm for human action recognition on raw depth maps. At the beginning, for each class we train a separate one-against-all convolutional neural network (CNN) to extract class-specific features representing person shape. Each class-specific, multivariate time-series is processed by a Siamese multichannel 1D CNN or a multichannel 1D CNN to determine features representing actions. Afterwards, for the nonzero pixels representing the person shape in each depth map we calculate statistical features. On multivariate time-series of such features we determine Dynamic Time Warping (DTW) features. They are determined on the basis of DTW distances between all training time-series. Finally, each class-specific feature vector is concatenated with the DTW feature vector. For each action category we train a multiclass classifier, which predicts probability distribution of class labels. From pool of such classifiers we select a number of classifiers such that an ensemble built on them achieves the best classification accuracy. Action recognition is performed by a soft voting ensemble that averages distributions calculated by such classifiers with the largest discriminative power. We demonstrate experimentally that on MSR-Action3D and UTD-MHAD datasets the proposed algorithm attains promising results and outperforms several state-of-the-art depth-based algorithms.

computer science, artificial intelligence