Graph-Based Gaze Estimation Without Fixed Head Pose Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gaze estimation methods require a fixed relative pose between the gaze estimation apparatus and the user's head, making it difficult to maintain for extended periods and leading to inaccurate gaze estimation.
Innovation Solution
A gaze estimation method using a graph representation-based approach, where gaze feature points are determined from eye data, a graph model is constructed, and gaze data is calculated through a pre-trained graph machine learning model, allowing for accurate estimation without the need for calibration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a fixed relative pose between the gaze estimation apparatus and the user's head is required, then measurement precision is improved, but ease of operation deteriorates
Solution Approach 1:
The patent applies dynamics by making the gaze estimation system adaptable to changing relative poses between the apparatus and user's head. The system dynamically adjusts to different positions and orientations without requiring the user to maintain a fixed pose, thereby resolving the contradiction between measurement precision and ease of operation.
Solution Approach 2:
The patent changes the parameters of the gaze estimation algorithm to accommodate variations in relative pose. By adjusting estimation parameters based on detected pose changes, the system maintains accuracy across different operational conditions, resolving the trade-off between precision and operational ease.
2Measurement precision
If a gaze calibration process is implemented, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent applies preliminary action by pre-training the gaze estimation model during device manufacturing or initial setup. This preliminary calibration phase creates a robust model that can operate accurately without requiring real-time calibration, thus resolving the contradiction between measurement precision and time loss by performing the calibration work beforehand.
Solution Approach 2:
The system applies self-service through automatic gaze estimation that requires no manual calibration input from the user. The model autonomously adapts to individual users through the pre-trained graph machine learning approach, eliminating the need for time-consuming calibration procedures while maintaining high accuracy.
3Measurement precision
If traditional gaze estimation methods are used, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent replaces traditional mechanical or geometric gaze estimation methods with a graph machine learning model. This substitution of the estimation mechanism enables higher precision by leveraging learned patterns from training data, while the graph-based approach provides a structured framework that manages the increased complexity in a systematic way.
Solution Approach 2:
The patent changes the fundamental parameters of the estimation system by transitioning from simple geometric calculations to a learned model with multiple parameters. The graph machine learning model incorporates numerous parameters that capture complex eye movement patterns, achieving superior precision despite the increased complexity through systematic parameter management.
Data Source
AI summary
The present invention provides a gaze estimation method and apparatus, a readable storage medium, and an electronic device. The method includes: acquiring eye data and determining state and position information of multiple gaze feature points based on the eye data; using each of the gaze feature points as a node, establishing a relationship between the nodes to obtain a graph model; determining feature information of the graph model based on the state and position information of each of the gaze feature points, and assigning the feature information to the graph model to obtain a graph representation corresponding to the eye data; and inputting the graph representation into a graph machine learning model, performing gaze estimation through the graph machine learning model, and outputting gaze data.