Eye-Tracking Authentication for Real-Time Deepfake Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing deepfake detection technologies struggle with real-time, continuous authentication of users, particularly in virtual environments, and lack effective age verification methods, making them susceptible to synthetic identities and underage access.
Innovation Solution
Utilizing AI models trained on eye-tracking data to analyze image frames for real-time authentication, including features like Fixation Duration, Saccade, Pupil Diameter, and Blink Rate, to determine user identity and age, and adjust device states based on similarity comparisons.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If iris scanning is used for deepfake detection, then detection accuracy is improved, but user workflow is blocked and authentication cannot be performed in real time
Solution Approach 1:
The patent changes the measurement parameters from static iris patterns to dynamic eye-tracking characteristics (fixation duration, saccade patterns, pupil diameter changes, blink rate). These dynamic parameters enable continuous authentication during normal user interaction without requiring the user to stop their workflow for specialized scanning.
Solution Approach 2:
The system transitions from static iris scanning to dynamic eye-tracking analysis that captures temporal variations in eye behavior. By analyzing how eye parameters change over time during natural interaction, the system achieves both accurate deepfake detection and seamless user experience.
2Reliability
If advanced machine learning models with large datasets are used for deepfake detection, then detection capability is improved, but computational complexity and processing time increase
Solution Approach 1:
The patent extracts and focuses on specific, discriminative eye-tracking parameters (fixation duration, saccade patterns, pupil diameter, blink rate) rather than using comprehensive deep learning models. This selective extraction of key features reduces computational complexity while maintaining detection effectiveness.
Solution Approach 2:
Instead of applying full-scale deep learning analysis to all image data, the system performs partial analysis focused specifically on eye-region characteristics and their temporal dynamics. This targeted approach achieves sufficient detection capability with reduced computational burden.
3Ease of manufacture
If manual age verification processes are used, then implementation simplicity is maintained, but verification reliability and security are compromised
Solution Approach 1:
The system automatically performs age verification by analyzing eye-tracking characteristics without requiring manual user input or intervention. The eye parameters naturally correlate with age groups, enabling the system to self-determine age category while maintaining implementation simplicity.
Solution Approach 2:
The patent replaces manual self-declaration of age with automated biometric analysis based on eye-tracking data. This substitution of mechanical/manual verification with optical-biometric analysis significantly improves verification reliability while maintaining ease of implementation through existing camera and processing capabilities.
Data Source
AI summary
Embodiments relate to a computer-implemented method for identification of image characteristics. The method includes analysing the contents of a plurality of image frames using one or more artificial intelligence (AI) models trained on historical data including human eye-tracking data. The analysing includes identifying one or more characteristics associated with eye-tracking data of a subject in the image frames, and comparing the one or more characteristics with historical data on which the AI models are trained to determine a degree of similarity between the characteristics and the historical data. The state of a first user electronic device or a second electronic device is changed based on a result of the comparison of characteristics with the historical data. Embodiments also relate to machine-readable information printed on a lens of the first electronic device or a contact lens in conjunction with the eye-tracking data analysis, as an additional security layer.