L_HAR : Lightweight Human Activity Recognition
Inho-Gil
January 02, 2025
Goal
To implement and lighten HAR with minimal imu setup.
Summary
This study aims to increase the efficiency of lightweight with on-device efficiency such as model size, number of parameters, and latency, rather than implementing accuracy in imu-based HAR. Therefore, we compare the lightweight results when applying several deep learning models (TCN, LSTM, Transformer).
