Related Experiment Video
Updated: Aug 19, 2025
03:56
A View of Their Own: Capturing the Egocentric View of Infants and Toddlers with Head-Mounted Cameras
Published on: October 5, 2018
7.5K
Detecting Screen Presence with Activity-Oriented RGB Camera in Egocentric Videos
Amit Adate1, Soroush Shahi2,3, Rawan Alharbi2,3
1Department of Electrical and Computer Engineering, Northwestern University, Evanston, IL, USA.
Proceedings of the ... IEEE International Conference on Pervasive Computing and Communications Workshops : Percom ... IEEE International Conference on Pervasive Computing and Communications. Workshops
|November 30, 2022
Summary
This study introduces a new deep learning method to detect screen time using low-resolution wearable cameras. This technology can help understand screen time
Area of Science:
- Computer Vision
- Human-Computer Interaction
- Digital Health
Background:
- Screen time is linked to health risks like mindless eating and poor academic performance.
- Traditional self-report measures for screen time are unreliable.
- Existing automatic detection methods often use high-resolution video, posing privacy concerns.
Purpose of the Study:
- To develop a deep learning method for detecting screen presence using low-resolution, activity-oriented wearable cameras.
- To address privacy concerns associated with traditional screen time assessment.
- To enable accurate, long-term monitoring of screen time in health studies.
Main Methods:
- Utilized deep learning algorithms on low-resolution RGB frames from activity-oriented wearable cameras.
- Tested the system on data from 10 individuals over 80 hours, analyzing 1.2 million frames.
- Focused on detecting various screen types (TVs, smartphones, laptops, tablets) with variable pixel density.
Main Results:
- Achieved an 81% F1-score, outperforming current state-of-the-art video screen detection methods.
- Demonstrated the feasibility of detecting screens from activity-oriented cameras.
- Successfully identified screen presence despite challenges like camera orientation and low resolution.
Conclusions:
- Activity-oriented cameras offer a privacy-preserving approach for screen time monitoring.
- This method facilitates a deeper understanding of screen time's impact on health behaviors.
- The developed system holds potential for large-scale, longitudinal health research.
