Lumina AI: Real-Time Eye Wellness Monitoring Using Advanced Computer Vision and Machine Learning

Abstract

Digital eye strain, clinically known as Computer Vision Syndrome (CVS), affects approximately 69% of digital device users, creating a significant public health challenge in the modern workplace. Traditional wellness monitoring systems lack personalization, medical integration, and intelligent detection capabilities. This study presents Lumina AI, a cross-platform desktop application that integrates real-time eye detection with intelligent wellness monitoring through advanced computer vision and machine learning algorithms. The system employs a hybrid architecture combining an Electron-based frontend with a Python-powered computer vision backend for optimal user experience and performance. Using OpenCV and dlib for 68-point facial landmark detection, the application calculates Eye Aspect Ratio (EAR) to precisely monitor blink frequency with 85.4% accuracy. A multi-modal reminder engine provides adaptive visual, auditory, and contextual interventions while supporting medical conditions such as Meibomian Gland Dysfunction (MGD). The personalization module customizes reminder intervals, display preferences, and therapeutic schedules according to individual user requirements. Experimental evaluation demonstrates lightweight performance (82.6 MB memory usage, 3.2% CPU consumption) and robust reliability across diverse environmental conditions. Lumina AI successfully bridges the gap between clinical eye health research and practical wellness applications, offering a scalable solution for integration with workplace health programs and digital wellness initiatives.

Introduction

The widespread use of digital devices has led to increased screen time, causing Computer Vision Syndrome (CVS), which affects approximately 66–69% of users and results in symptoms such as dry eyes, visual fatigue, and reduced concentration. Traditional preventive measures, like timer-based reminders, fail to account for individual physiological patterns, context, and medical needs, while existing commercial solutions are fragmented, platform-limited, and often compromise privacy.

This study introduces Lumina AI, a desktop application designed to address these gaps. Key innovations include 68-point facial landmark detection for precise blink monitoring, integration of Meibomian Gland Dysfunction (MGD) therapeutic protocols, privacy-first local processing, cross-platform support for Windows and macOS, and adaptive machine learning-based personalization of interventions.

The literature review highlights various eye blink detection techniques—ranging from motion-based and image-based methods to deep learning and capacitive sensing—and notes limitations such as sensitivity to lighting, head movements, computational complexity, and lack of integration with medical protocols. Commercial solutions, such as EyeLeo, AutoBlink, and SightKick AI, show limited personalization, platform restrictions, or privacy concerns. Lumina AI addresses these gaps with real-time, offline, clinically informed monitoring, providing an innovative tool for occupational eye health management.

Conclusion

This study presented Lumina AI, a comprehensive desktop application for real-time eye wellness monitoring that addresses critical gaps in current digital health solutions. Through the integration of advanced computer vision algorithms, medical protocol support, and privacy-preserving architecture, the system achieves superior performance compared to existing alternatives. Key contributions include the first consumer implementation of 68-point facial landmark detection for blink monitoring, integration of clinical protocols for MGD support, and demonstration of effective cross-platform deployment. Experimental validation shows 85.4% detection accuracy with minimal system resource requirements, confirming the feasibility of real-time eye health monitoring on standard computing hardware. The system\'s modular architecture and privacy-first design position it for scalable deployment in workplace environments while supporting individual medical needs. Future development will focus on expanding therapeutic protocol support, improving detection algorithms through advanced machine learning, and enabling integration with emerging healthcare technologies. Lumina AI represents a significant step toward intelligent, personalized digital wellness solutions that bridge the gap between clinical research and practical application in everyday computing environments.

References

[1] Computer Vision Syndrome Research Group, \"Digital Eye Strain: A Comprehensive Analysis of Prevalence and Impact,\" Journal of Occupational Health, vol. 65, no. 3, pp. 178-192, 2023. [2] American Optometric Association, \"Computer Vision Syndrome: Clinical Guidelines and Treatment Protocols,\" Optometry Today, vol. 61, no. 8, pp. 45-58, 2022. [3] T. Soukupová and J. ?ech, \"Eye blink detection using facial landmarks,\" in Proc. 21st Computer Vision Winter Workshop, Feb. 2016, pp. 1-8. [4] S. Kumar and R. Patel, \"Automated eye strain detection in digital environments: A systematic review,\" IEEE Trans. Biomed. Eng., vol. 68, no. 12, pp. 3401-3412, 2021. [5] M. Johnson et al., \"Privacy-preserving health monitoring applications: Design principles and implementation challenges,\" ACM Computing Surveys, vol. 54, no. 7, pp. 1-35, 2022. [6] International Eye Health Consortium, \"Global Prevalence of Computer Vision Syndrome: Meta-Analysis of Cross-Sectional Studies,\" Ophthalmology Research, vol. 89, no. 4, pp. 234-248, 2023. [7] A. Drutarovsky and A. Fogelton, \"Eye blink detection using variance of motion vectors,\" in Proc. European Conference on Computer Vision, Zurich, Switzerland, Sep. 2014, pp. 436-448. [8] F. Song, X. Tan, X. Liu, and S. Chen, “A multi-scale histogram-based analysis employing principal oriented gradient features integrated with machine learning classifiers,” Pattern Recognition, vol. 47, no. 9, pp. 2825–2838, 2014. [9] M. Chau and M. Betke, \"Real time eye tracking and blink detection with USB cameras,\" Computer Science Technical Report, Boston University, Boston, MA, USA, Tech. Rep. BU-CS-TR-2005-12, 2005. [10] P. Viola and M. Jones, \"Rapid object detection using a boosted cascade of simple features,\" in Proc. IEEE Computer Vision and Pattern Recognition, Kauai, HI, USA, Dec. 2001, pp. 511-518. [11] X. Liu, Y. Wang, and Z. Chen, “A capacitive sensing approach for non-contact eye-blink monitoring with enhanced energy efficiency and low latency,” IEEE Sensors Journal, 2024. [12] S. Kim et al., \"Real-time driver monitoring system with facial landmark detection,\" in Proc. IEEE Conf. Computer Vision and Pattern Recognition, 2023. [13] L. Hong et al., “Dual-embedding transformer network for reliable eye-blink recognition in video data,” in Proceedings of the Computer Vision Conference, 2024. [14] Y. Xiong et al., \"A review of deep learning in blink detection,\" IEEE Review, 2025. [15] A. Di Nisio et al., \"Noise robustness evaluation of image processing algorithms for eye blinking detection,\" IEEE Trans. Instrumentation and Measurement, 2025. [16] F. Attivissimo et al., \"Performance evaluation of image processing algorithms for eye blink detection,\" Computers & Vision, vol. 2023, 2023. [17] A. Zhang and B. Liu, \"Real-time eye tracking using Haar cascade classifiers,\" in Proc. IEEE Conf. Computer Vision and Pattern Recognition, pp. 892-899, 2019. [18] Digital Wellness Technologies Inc., \"Comparative Analysis of Eye Health Applications,\" Tech. Report TDW-2023-01, 2023. [19] AI Health Solutions, \"Cloud-Based Eye Monitoring: Privacy and Performance Trade-offs,\" in Proc. Digital Health Conf., pp. 78-85, 2023.

Copyright

Copyright © 2025 Usha Dhankar, Ayush Kumar, Nandan Pradhan, Nishant Gupta. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.