Real-time human activity recognition from accelerometer data using Convolutional Neural Networks

• Published in: Applied soft computing Vol. 62; pp. 915 - 922
• Main Author: Ignatov, Andrey
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.01.2018
• Subjects: Activity recognition; Convolutional Neural Networks; Deep learning; Feature extraction; Time series classification; Deep learning; Activity recognition; Feature extraction; Convolutional Neural Networks; Time series classification
• ISSN: 1568-4946; 1872-9681
• DOI: 10.1016/j.asoc.2017.09.027

[Display omitted] •We combine a shallow CNN for local feature extraction with statistical features.•We study how time series length affects the recognition accuracy.•Limit time series length up to 1s to enable real-time activity classification.•Limit time series length up to 1s to enable real-time activity classification.•Perform a cross-dataset evaluation to ensure model user- and platform-independency.•Test the solution on desktop and mobile devices to guarantee acceptable running time.•We make the source code of the model and the whole pipeline publicly available. With a widespread of various sensors embedded in mobile devices, the analysis of human daily activities becomes more common and straightforward. This task now arises in a range of applications such as healthcare monitoring, fitness tracking or user-adaptive systems, where a general model capable of instantaneous activity recognition of an arbitrary user is needed. In this paper, we present a user-independent deep learning-based approach for online human activity classification. We propose using Convolutional Neural Networks for local feature extraction together with simple statistical features that preserve information about the global form of time series. Furthermore, we investigate the impact of time series length on the recognition accuracy and limit it up to 1s that makes possible continuous real-time activity classification. The accuracy of the proposed approach is evaluated on two commonly used WISDM and UCI datasets that contain labeled accelerometer data from 36 and 30 users respectively, and in cross-dataset experiment. The results show that the proposed model demonstrates state-of-the-art performance while requiring low computational cost and no manual feature engineering.