Learning Attribute Representation for Human Activity Recognition

TL;DR

This paper introduces deep architectures to learn attribute representations from sensor data for human activity recognition, addressing the lack of labeled attributes and outperforming state-of-the-art methods.

Contribution

It proposes novel deep models for attribute prediction in sensor-based activity recognition, filling the gap of labeled attributes in this domain.

Findings

Deep architectures outperform existing methods

Learned attribute representations improve recognition accuracy

Models generalize well across datasets

Abstract

Attribute representations became relevant in image recognition and word spotting, providing support under the presence of unbalance and disjoint datasets. However, for human activity recognition using sequential data from on-body sensors, human-labeled attributes are lacking. This paper introduces a search for attributes that represent favorably signal segments for recognizing human activities. It presents three deep architectures, including temporal-convolutions and an IMU centered design, for predicting attributes. An empiric evaluation of random and learned attribute representations, and as well as the networks is carried out on two datasets, outperforming the state-of-the art.