Benchmark on Activity Recognition based on Wearable Sensors

Benchmark on Activity Recognition based on Wearable Sensors

This page contains the source code and data used in our paper “Human Activity Recognition Based on Wearable Sensor Data: A Standardization of the State-of-the-Art“. In this paper, we implement and evaluate several state-of-the-art approaches, ranging from methods based on handcrafted features to convolutional neural networks. Also, we standardize a large number of datasets, which vary in terms of sampling rate, number of sensors, activities, and subjects. We implement and evaluate several state-of-the-art methods, summarized in the table below.

Results

Mean accuracy achieved by the methods using the Leave-One-Subject-Out (LOSO) as validation protocol. The symbol ‘x’ means that it was not possible to execute the method on the respective dataset.
Method MHEALTH PAMAP2 USCHAD UTD-1 UTD-2 WHARF WISDM
Kwapisz et al. 90.41 71.27 70.15 13.04 66.67 42.19 75.31
Catal et al. 94.66 85.25 75.89 32.45 74.67 46.84 74.96
Kim et al. 93.90 81.57 64.20 38.05 64.60 51.48 50.22
Chen and Xue 88.67 83.06 75.58 x x 61.94 83.89
Jiang and Yin 51.46 x 74.88 x x 65.35 79.97
Ha et al. 88.34 73.79 x x x x x
Ha and Choi 84.23 74.21 x x x x x

Download

Download the code from GitHub.

Reference

If you use the code or parts of it, please cite the following paper.

1.Jordão, Artur; Jr., Antonio C. Nazare; Sena, Jessica; Schwartz, William Robson (2018): Human Activity Recognition Based on Wearable A Standardization of the state-of-the-art. In: arXiv, pp. 1-12, 2018. (Type: Article | Links | BibTeX)