What is it about?
The area of human activity recognition by mobile phone sensors is still unresolved and an accurate method is not developed yet. In our paper, we proposed and applied a new method of human activity detection using mobile phone's GPS and Wi-fi modules. We used the Deep learning method to detect automatically features of the activities which will maximize differences among different activities. As part of our research, we also analyzed generated features. As a final result, we achieved an accuracy of approximately ~80% on 8 activities (still, walk, run, bike, car, bus, train, subway).
Featured Image
Photo by Alexandre Boucher on Unsplash
Why is it important?
The automatic and accurate method of human activity detection by mobile phone is the base of the modern security and healthcare technologies for mobile phones (such as phone theft detection, continuous authentication, user's health tracking, etc.)
Perspectives
I hope this paper will become the basis of accurate human activity recognition technologies. Because the absence of such technologies greatly inhibits the development of new technologies in the mobile phones sphere.
Livii Iabanzhi
Taras Shevchenko National University of Kyiv
Read the Original
This page is a summary of: Location-based Human Activity Recognition Using Long-term Deep Learning Invariant Mapping, September 2021, ACM (Association for Computing Machinery),
DOI: 10.1145/3460418.3479381.
You can read the full text:
Resources
Contributors
The following have contributed to this page
