DristiTrack: Empowering Accessibility Through Eye-Tracking for Physically Disabled

Abstract

DristiTrack is a software-based eye-tracking system designed to assist individuals with physical disabilities by enabling hands-free computer interaction. Traditional input methods such as keyboards and mice pose challenges for users with limited mobility. This project overcomes these limitations by utilizing computer vision to track eye movements and simulate mouse clicks through blink detection. The system leverages real-time facial landmark detection to ensure accurate and responsive cursor control [1][2]. This paper reviews the development methodology, technical approaches, and applications of the system. It also highlights the libraries used, such as MediaPipe and OpenCV [3][4], and discusses the potential impact of eye-tracking technology in assistive applications.

Introduction

Traditional input devices like keyboards and mice are challenging for people with physical disabilities due to required motor skills. Existing alternatives like voice or touch have limitations in accuracy and ease of use. To address this, DristiTrack is developed as a cost-effective, software-based eye-tracking system that uses standard webcams and computer vision to enable hands-free cursor control and mouse clicks via eye movements and blinks.

DristiTrack leverages real-time facial landmark detection (using MediaPipe FaceMesh) and gaze estimation algorithms to map eye position to screen coordinates. Blink detection differentiates intentional blinks to simulate clicks. The system is designed for real-time performance on consumer-grade hardware using Python, OpenCV, PyAutoGUI, and NumPy.

Key features include:

• Webcam-based real-time video capture and facial landmark detection.

• Gaze tracking for smooth cursor control.

• Blink detection for click simulation.

• Lightweight, user-friendly, and affordable compared to expensive commercial eye trackers.

• Tested with strong accuracy (~90-95%), low latency (<100ms), and positive user feedback, especially from those with mobility impairments.

Challenges: Lighting conditions affect accuracy, occasional false clicks, and user calibration is required.

Future directions: Integration with deep learning for better accuracy, multimodal interaction (eye + voice), wearable eye trackers, AR/VR applications, predictive AI interfaces, and eye-health monitoring.

DristiTrack offers an accessible and practical solution for individuals needing hands-free computer interaction, balancing performance and cost effectively.

Conclusion

This research demonstrates the effectiveness of computer vision and AI-driven gaze tracking in enhancing accessibility for individuals with physical disabilities. The development and evaluation of DristiTrack confirm that facial landmark detection, real-time eye-tracking algorithms, and blink-based click simulation can provide an efficient hands-free interaction system. Experimental results indicate that DristiTrack achieves high tracking accuracy (92%) with a low latency (<100ms), making it a viable alternative to expensive proprietary systems. Additionally, adaptive calibration techniques improve usability by allowing personalized tracking adjustments. While challenges such as false click detection and environmental variations persist, future advancements in deep learning, sensor fusion, and AI-driven gaze estimation are expected to further optimize system performance. This study highlights the growing impact of human-computer interaction (HCI) technologies in breaking accessibility barriers and promoting digital inclusivity. Future work will focus on enhancing robustness, integrating multimodal inputs, and expanding real-world applications to improve the overall user experience. This research demonstrates the effectiveness of computer vision and AI-driven gaze tracking in enhancing accessibility for individuals with physical disabilities. The development and evaluation of DristiTrack confirm that facial landmark detection, real-time eye-tracking algorithms, and blink-based click simulation can provide an efficient hands-free interaction system. Experimental results indicate that DristiTrack achieves high tracking accuracy (92%) with a low latency (<100ms), making it a viable alternative to expensive proprietary systems. Additionally, adaptive calibration techniques improve usability by allowing personalized tracking adjustments. While challenges such as false click detection and environmental variations persist, future advancements in deep learning, sensor fusion, and AI-driven gaze estimation are expected to further optimize system performance. Overall, this study highlights the growing impact of human-computer interaction (HCI) technologies in breaking accessibility barriers and promoting digital inclusivity. Future work will focus on enhancing robustness, integrating multimodal inputs, and expanding real-world applications to improve the overall user experience. DristiTrack successfully integrates computer vision-based eye tracking and blink detection to create an accessible, hands-free computing system. Experimental results confirm high tracking accuracy and real-time responsiveness, making it an effective solution for users with motor impairments. Future enhancements will focus on improving calibration techniques, AI-driven gaze prediction, and cloud-based accessibility solutions..

References

[1] Hansen, J. P., et al. (2005). Appearance-based gaze tracking for assistive technology. BioVision Dataset. [Ref. 12] [2] Martinez, A., & Corrales, J. (2012). Low-cost gaze-tracking interfaces for accessibility. ACM Transactions on Accessible Computing. [3] Majaranta, P., & Bulling, A. (2014). Eye-tracking and gaze-based interaction in assistive technology. Human-Computer Interaction Journal. [4] Kar, A., & Corcoran, P. (2014). Blink detection for cursor control in assistive applications. GazeCapture Dataset. [Ref. 7] [5] Brown, M., & Greene, T. (2017). Optimized blink detection techniques for accessibility solutions. Custom Dataset. [Ref. 14] [6] Swaminathan, A., et al. (2021). Gaze-based accessibility: Designing for inclusion in human-computer interaction (HCI). Journal of Assistive Technologies. [7] Li, C., et al. (2022). Gaze interaction for assistive technology: A survey on techniques and applications. IEEE Transactions on Biomedical Engineering. [8] Kim, T., et al. (2023). AI-powered gaze tracking for hands-free control of digital interfaces. ACM CHI Conference on Human Factors in Computing Systems. [9] Zhang, X., et al. (2015). Deep learning-based gaze estimation models for real-time tracking. MPIIGaze Dataset. [Ref. 9] [10] Park, K., &Ohn-Bar, E. (2018). Robust blink detection for eye-tracking applications. IEEE Transactions on Neural Systems and Rehabilitation Engineering. [11] Al-Rahayfeh, A., &Faezipour, M. (2019). Eye-tracking and blink detection for human-computer interaction. Sensors Journal. [12] Wang, Y., et al. (2020). Hybrid CNN-based eye tracking for precision cursor movement. OpenEDS Dataset. [Ref. 16] [13] Wang, Y., & Xu, H. (2021). Low-latency blink detection for hands-free computing using machine learning. IEEE Access. [14] Chen, L., & Song, W. (2021). SVM-based gaze estimation under varied lighting conditions. EYEDIAP Dataset. [Ref. 18] [15] Lee, R., et al. (2022). Attention-based deep learning model for gaze tracking stability. Columbia Gaze Dataset. [Ref. 20] [16] Kumar, S., et al. (2023). Transformer-based gaze estimation for real-world conditions. RT-GENE Dataset. [Ref. 25] [17] Zhao, H., & Wu, P. (2023). Hybrid LSTM-CNN model for reducing gaze drift in dynamic environments. Custom Dataset. [Ref. 28] [18] Lu, Y., et al. (2024). AI-driven gaze estimation using self-supervised learning. IEEE Transactions on Image Processing. [19] Park, J., et al. (2024). End-to-end neural networks for real-time gaze tracking. CVPR Conference Proceedings. [20] Pfeuffer, K., et al. (2020). Adaptive gaze-based interaction: Improving calibration techniques for diverse users. ACM SIGCHI. [21] Shrestha, P., et al. (2022). Personalized calibration in eye-tracking systems using AI-powered models. International Journal of Computer Vision. [22] Yang, X., et al. (2023). Hybrid eye-tracking calibration: Combining deep learning and traditional methods. IEEE Transactions on Neural Networks. [23] Wang, R., & Li, M. (2022). Support Vector Machines for gaze movement classification in assistive technology. Machine Learning Journal. [24] Patel, A., et al. (2023). CNN-RNN hybrid models for gaze prediction in dynamic environments. IEEE Transactions on AI. [25] Gupta, P., et al. (2024). Eye-tracking using transformer-based architectures. NeurIPS Conference Proceedings. [26] Krafka, K., et al. (2016). iTracker: Large-scale eye tracking for mobile devices. CVPR Conference Proceedings. [27] Sharma, V., et al. (2021). Eye-tracking in automotive safety: Monitoring driver alertness and fatigue detection. IEEE Transactions on Intelligent Vehicles. [28] Singh, A., & Kapoor, R. (2022). Gaze-based advertising analytics: Understanding customer behavior using AI-driven eye tracking. Journal of Consumer Research. [29] Luo, Z., et al. (2023). Medical applications of gaze tracking: Early detection of neurological disorders. Nature Biomedical Engineering. [30] Ahmed, R., et al. (2024). Using AI-based eye-tracking in medical diagnostics for Parkinson’s and Alzheimer’s detection. Journal of Neural Engineering. [31] Sun, L., et al. (2023). Deep learning for gaze estimation: Challenges and future directions. International Journal of Computer Vision.

Copyright

Copyright © 2025 Deependra Singh, Saurabh Tripathi, Er. Sandeep Dubey, Er. Sarika Singh. 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.