On-the-Fly Eye Tracking Calibration from Dwell Actions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing eye tracking systems on mobile devices require burdensome calibration processes that consume significant computational resources and interrupt user experience, necessitating a hands-free, on-the-fly calibration technique to improve accuracy while reducing user burden.
Innovation Solution
A method for on-device eye tracking that utilizes a dwelling action on user interface components to collect calibration pairs, leveraging a regression model for real-time calibration without requiring additional input confirmation, using stimuli marks overlaid on the interface to obtain gaze information and train a calibration model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional calibration methods are used to map eye signals to screen, then calibration accuracy can be improved by collecting more valid calibration points, but the user burden and computational resource consumption increase significantly
Solution Approach 1:
The system performs calibration automatically without requiring active user participation. The calibration process leverages existing user interactions with the interface, where the system autonomously collects calibration data by detecting dwell actions and mapping them to calibration points, thereby eliminating user burden while maintaining calibration accuracy
Solution Approach 2:
The calibration process occurs continuously in the background during normal user interaction rather than requiring discrete calibration sessions. The system continuously collects calibration pairs from dwell actions, updating the calibration model progressively, which maintains accuracy without interrupting user experience
2Reliability
If more calibration points are collected to improve gaze estimation reliability, then measurement precision improves, but computational resources and time consumption increase
Solution Approach 1:
The system collects calibration data partially by focusing on high-quality calibration pairs derived from dwell actions, which are naturally occurring user interactions. This selective collection approach obtains sufficient calibration points for reliable gaze estimation without the excessive computational cost of processing all possible user interactions
Solution Approach 2:
The system uses feedback from dwell action detection to identify and collect only the most relevant calibration pairs. By monitoring user interaction patterns and detecting when dwell actions occur, the system efficiently selects calibration data that maximizes reliability while minimizing computational resources
Data Source
AI summary
Calibration of eye tracking is improved by collecting additional calibration pairs when user is using apps with eye tracking. A user input component is presented on a display of an electronic device, detecting a dwelling action for user input component, and in response to detecting the dwelling action, obtaining a calibration pair comprising an uncalibrated gaze point and a screen location of the user input component, wherein the uncalibrated gaze point is determined based on an eye pose during the dwelling action. A screen gaze estimation is determine based on the uncalibrated gaze point, and in response to determining that the calibration pair is a valid calibration pair, training a calibration model using the calibration pair.