Event-Based Gaze Tracking via Asynchronous Pixel Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional eye tracking systems face limitations in update frequency, power consumption, and form factor, making them less suitable for near-eye gaze tracking, as they require high bandwidth and sacrifice device portability due to frame-based operation principles.
Innovation Solution
Event-based gaze tracking systems utilize event cameras to asynchronously sample pixels, providing event data for inferring eye position and gaze, which can be updated rapidly, combined with frame data to correct inaccuracies, allowing for higher performance and real-time tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If frame-based eye tracking systems are used, then measurement precision is maintained, but update frequency is limited and power consumption increases
Solution Approach 1:
The system dynamically switches between frame-based and event-based operation modes. Frame-based mode provides periodic accurate gaze estimates, while event-based mode provides high-frequency updates during rapid eye movements. The system adapts its operation dynamically based on scene changes and tracking requirements, resolving the contradiction between precision and update frequency.
Solution Approach 2:
The system uses periodic frame capture at lower frequencies combined with continuous event-based sampling. Frames are captured at regular intervals to provide stable reference points, while events provide continuous high-frequency updates between frames. This periodic action allows the system to maintain precision while achieving high update frequencies during dynamic events.
2Measurement precision
If frame-based eye tracking systems are used, then gaze tracking accuracy is maintained, but power consumption increases
Solution Approach 1:
The system dynamically adjusts its power consumption based on tracking needs. During stable gaze periods, it relies on lower-power event-based sampling. During rapid eye movements or when high accuracy is required, it activates frame-based capture. This dynamic power management resolves the contradiction between maintaining accuracy and reducing power consumption.
Solution Approach 2:
The system discards redundant frame data when events provide sufficient tracking information, and recovers frame-based operation when higher precision is needed. By selectively using frame data only when necessary and relying on event data otherwise, the system reduces overall power consumption while maintaining accuracy requirements.
3Productivity
If high bandwidth eye tracking systems are used, then update rate is improved, but device portability is reduced
Solution Approach 1:
The system dynamically adjusts its bandwidth requirements based on scene activity. During static scenes, it operates at low bandwidth using event data alone. During rapid eye movements, it temporarily increases bandwidth by capturing frames. This dynamic bandwidth adjustment allows high update rates when needed while maintaining portability through reduced average bandwidth requirements.
Solution Approach 2:
The system uses periodic frame capture at lower bandwidth intervals combined with continuous event-based tracking. This periodic action allows the system to achieve high effective update rates during dynamic events while maintaining low average bandwidth consumption, thus preserving device portability.
4Productivity
If event-based sampling is used, then update frequency is improved, but measurement precision during rapid movements may degrade
Solution Approach 1:
The system uses frame data as an intermediary to correct and refine event-based gaze estimates during rapid eye movements. Events provide high-frequency updates, while periodic frames serve as reference points to calibrate and correct drift. This intermediary role of frame data resolves the contradiction by maintaining precision during rapid movements while preserving high update frequencies.
Solution Approach 2:
The system implements feedback by using frame-based gaze estimates to correct and refine event-based tracking during rapid eye movements. The frame data provides feedback on overall gaze position accuracy, which is then used to adjust and correct the high-frequency event-based estimates, maintaining precision while achieving high update rates.
Data Source
AI summary
Systems and methods for event-based gaze in accordance with embodiments of the invention are illustrated. One embodiment includes an event-based gaze tracking system, including a camera positioned to observe an eye, where the camera is configured to asynchronously sample a plurality of pixels to obtain event data indicating changes in local contrast at each pixel in the plurality of pixels, and a processor communicatively coupled to the camera, and a memory communicatively coupled to processor, where the memory contains a gaze tracking application, where the gaze tracking application directs the processor to, receive the event data from the camera, fit an eye model to the eye using the event data, map eye model parameters from the eye model to a gaze vector, and provide the gaze vector.