Unleashing the Power of Shared Label Structures for Human Activity Recognition

By X. Zhang et al.

Published on June 10, 2023

Read the original document by opening this link in a new tab.

Table of Contents

1. INTRODUCTION
2. RELATED WORK
3. PRELIMINARY
4. METHODOLOGY

Summary

The paper discusses the SHARE framework for Human Activity Recognition, focusing on modeling label structures to enhance knowledge sharing across activities. The framework consists of a Time-Series Encoder and a Label Structure-Constrained Decoder, leveraging shared structures in label names to improve recognition performance. Three augmentation methods are proposed to capture shared semantic structures more effectively.