Event-Based Eye Tracking With Glint-Guided Gaze Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional frame-based cameras for eye tracking are slow and produce large volumes of data, leading to high latency and inefficiency in processing eye movement data.
Innovation Solution
An eye tracking device utilizing an event-based optical sensor and a controller that processes a signal stream of events from glint sources to predict gaze information using glint information, reducing latency by generating predicted gaze information at higher frequencies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If an event-based optical sensor is used to capture eye movement data, then the tracking speed and processing efficiency are improved, but the data volume and complexity of processing increase
Solution Approach 1:
The patent extracts only the essential glint information from the event-based sensor data stream, separating the critical gaze-related signals from the surrounding noise and irrelevant events. This selective extraction reduces processing complexity while maintaining tracking speed advantages.
Solution Approach 2:
The system performs preliminary processing of event-based sensor data by accumulating events into intensity images before feeding them to the neural network. This pre-processing step organizes the high-volume event data into a more manageable format, reducing the computational burden during actual gaze prediction.
2Loss of time
If glint information is used to predict gaze information, then the latency is reduced and processing efficiency is improved, but the reliability of tracking may be compromised
Solution Approach 1:
The patent implements a feedback mechanism where the neural network continuously refines gaze predictions based on incoming glint information. The system uses the predicted gaze information to adjust and improve subsequent predictions, maintaining reliability even with reduced latency.
Solution Approach 2:
The system dynamically adjusts processing parameters based on the quality and quantity of available glint information. When glint data is sufficient, the system operates at high speed with minimal latency; when data quality deteriorates, the system automatically adjusts to maintain tracking reliability.
3Measurement precision
If a neural network is used to process event-based sensor data, then the accuracy of gaze prediction is improved, but the processing time and computational resources increase
Solution Approach 1:
The patent applies partial processing by using a simplified neural network architecture that processes only the most critical features of the event-based data. Instead of analyzing all possible parameters, the system focuses on the essential glint characteristics needed for accurate gaze prediction, reducing computational overhead.
Solution Approach 2:
The system performs preliminary accumulation of events into intensity images before neural network processing, pre-organizing the data in a format that maximizes neural network efficiency. This pre-processing reduces the actual computation time required during gaze prediction.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The device achieves faster and more reliable eye movement tracking by utilizing glint information to predict gaze information, reducing latency and improving processing efficiency.
Implementation Method 1
An event-based optical sensor produces a signal stream of events in response to radiation reflected off an eye of a user
Implementation Method 2
The eye tracking device comprises one or more glint sources. The one or more glint sources are configured to send radiation to the eye of the user
Data Source
AI summary
An eye tracking device that includes: an event-based optical sensor configured to receive radiation reflected off an eye of a user and produce a signal stream of events, each event corresponding to detection of a temporal change in the received radiation at pixels of the sensor; glint sources configured to send radiation to the eye that is reflected off the cornea and received by the sensor in the form of individual glints; a controller connected to the sensor and configured to receive the signal stream from the sensor and to generate: gaze information of the eye and first glint information of the glints, at a first instant of time; and second glint information of the glints at a second instant of time that is later than the first instant of time. The controller is configured to generate predicted gaze information at the second instant of time.