Personalized Gaze Estimation Calibration for Individual Eye Variation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing gaze estimation methods struggle with accuracy due to individual variations in the position of the fovea centralis, which is difficult to determine using captured images, leading to inconsistent gaze direction estimation across different individuals.

Innovation Solution

A gaze estimation apparatus that utilizes calibration information including feature and true value information to account for individual differences, using a learned estimation model with extractors and estimators to accurately estimate gaze direction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If the corneal reflection method is used to estimate gaze direction, then high accuracy is achieved, but the range of estimatable gaze directions is limited and processing cost increases

Engineering Contradiction:

Improvegaze direction estimation accuracyVSAvoidrange of estimatable gaze directions

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent creates a mapping relationship (copy) between pupil shape features and gaze directions obtained through corneal reflection method. Instead of directly using the complex corneal reflection method for all gaze directions, it copies the accurate gaze direction data to train a model that can generalize to directions beyond the calibration range, thus extending the estimatable range while maintaining accuracy.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The patent performs preliminary calibration by capturing images at multiple known gaze directions and establishing the mapping between pupil shapes and gaze directions before actual gaze estimation. This preliminary action creates a trained model that can then quickly estimate gaze directions without requiring real-time corneal reflection analysis, enabling both accuracy and extended range.

Inventive Principle:

Principle #10Preliminary action

2Productivity

If the pupil shape method is used to estimate gaze direction, then processing cost is reduced and speed increases, but accuracy decreases due to difficulty in accurately obtaining pupil shape

Engineering Contradiction:

Improvegaze estimation speedVSAvoidgaze direction estimation accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent uses the pupil shape method to extract features and copies the mapping relationship from the corneal reflection calibration data. By training a model with copied accurate gaze direction labels from calibration, the system maintains high accuracy while using the computationally efficient pupil shape extraction method for actual gaze estimation.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The patent changes the approach from directly measuring gaze direction through complex corneal reflection analysis to using pupil shape parameters (extracted via ellipse fitting) as input features to a trained model. This parameter change enables faster processing while the model training ensures accuracy is preserved through learning from calibrated data.

Inventive Principle:

Principle #35Parameter changes

3Adaptability or versatility

If a general gaze estimation model is used, then it can estimate gaze direction for any person, but accuracy decreases due to individual variations in fovea centralis position

Engineering Contradiction:

Improveapplicability to different individualsVSAvoidgaze direction estimation accuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent performs preliminary calibration for each individual by capturing images at multiple known gaze directions and training a personalized model. This preliminary action accounts for individual anatomical variations in the fovea centralis position, creating a customized mapping relationship that maintains high accuracy for that specific person while the overall system remains applicable to different individuals through repeated calibration.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent applies local quality by creating individual-specific calibration data and models rather than using a single universal model. Each person's unique eye characteristics (local quality) are captured through their own calibration process, allowing the system to adapt to individual variations in fovea centralis position while maintaining overall system versatility.

Inventive Principle:

Principle #3Local quality

Data Source

PatentEP4089628B1Sight line estimation device, sight line estimation method, model generation device, and model generation method

Publication Date: 2026.03.04 OMRON CORP

Figure 1
Figure 2~3
Figure 4A~4B

AI summary

A gaze estimation apparatus estimates the gaze direction of a target person using calibration information including feature information about the gaze of the eyes of the target person looking in a predetermined direction and true value information indicating the true value for the predetermined direction, in addition to a target image including the eye(s) of the target person. The gaze estimation apparatus can thus estimate the gaze direction of the target person reflecting individual differences, and allows estimation of the gaze direction of the target person with higher accuracy.