Abstract
When working from home, self-management becomes of paramount importance due to the absence of a boss or colleagues. As a result, individuals tend to waste time surfing the internet and playing with our smartphones. We propose a wrist-worn sensor-based system that identifies whether a desk worker is working or not for self-management and productivity. Our main hypothesis is that the identification of the various tasks that occur during desk work, such as using computers, reading books, manipulating a smartphone, and writing, can be simply distinguished by the direction of the palm. In this paper, to verify our hypothesis, we measure various tasks with the wrist-worn sensor attached to clarify the relationship between hand orientation and each task. At the same time, we develop a machine learning-based classifier to distinguish between the states of `working' and `not-working' using the obtained hand orientation data. We performed 10-fold cross-validation and Leave-One-Person-Out validation and we found that it was possible to distinguish whether a desk worker is working or not with an F1-value of 0.8 or higher.
Links
BibTeX (Download)
@inproceedings{Hayashi2021,
title = {Distinguishing Working State by Palm Orientation},
author = {Kenta Hayashi, Shingo Kumazoe, Shigemi Ishida, Yutaka Arakawa},
url = {https://ieeexplore.ieee.org/document/9391950},
doi = {10.1109/LifeTech52111.2021.9391950},
year = {2021},
date = {2021-03-10},
urldate = {2021-03-10},
booktitle = {2021 IEEE 3rd Global Conference on Life Sciences and Technologies (LifeTech 2021)},
abstract = {When working from home, self-management becomes of paramount importance due to the absence of a boss or colleagues. As a result, individuals tend to waste time surfing the internet and playing with our smartphones. We propose a wrist-worn sensor-based system that identifies whether a desk worker is working or not for self-management and productivity. Our main hypothesis is that the identification of the various tasks that occur during desk work, such as using computers, reading books, manipulating a smartphone, and writing, can be simply distinguished by the direction of the palm. In this paper, to verify our hypothesis, we measure various tasks with the wrist-worn sensor attached to clarify the relationship between hand orientation and each task. At the same time, we develop a machine learning-based classifier to distinguish between the states of `working' and `not-working' using the obtained hand orientation data. We performed 10-fold cross-validation and Leave-One-Person-Out validation and we found that it was possible to distinguish whether a desk worker is working or not with an F1-value of 0.8 or higher.},
keywords = {Activity recognition},
pubstate = {published},
tppubtype = {inproceedings}
}