Head-Wearable Eye Tracking With Low-Resolution Neural Gaze Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current head-mounted eye trackers are bulky, require high-resolution cameras, active infrared illumination, and complex computations, leading to discomfort, limited accuracy, and high power consumption, and are challenging to integrate into glasses frames due to occlusions and environmental light interference.
Innovation Solution
A head-wearable device with integrated cameras and a non-transparent component, using a neural network to determine gaze-related parameters without explicit feature extraction, allowing for improved accuracy and ergonomic design by implicitly accounting for camera pose variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-resolution cameras are used for eye tracking, then measurement precision is improved, but device complexity and weight increase
Solution Approach 1:
The patent changes the fundamental parameter of image resolution from high-resolution to low-resolution (e.g., 64x64 pixels or smaller). This parameter change enables the system to achieve acceptable gaze estimation accuracy without requiring complex high-resolution cameras, thereby reducing device complexity and weight while maintaining functionality.
2Measurement precision
If high-resolution cameras and active infrared illumination are used, then measurement precision is improved, but use of energy increases
Solution Approach 1:
The patent reduces the resolution parameter of the camera, which directly lowers the computational energy required for image processing. Additionally, the system eliminates or reduces active infrared illumination, further decreasing power consumption while maintaining sufficient accuracy for gaze estimation through alternative methods such as reflectance-based approaches.
3Device complexity
If cameras are integrated into glasses frames, then device complexity is reduced, but object-affected harmful factors increase due to occlusions and environmental light interference
Solution Approach 1:
The patent converts the harmful effect of environmental light and occlusions into a beneficial approach by using reflectance-based eye tracking. Instead of trying to eliminate environmental light interference, the system utilizes reflected light patterns from the eye surface (such as the corneal reflection) as key features for gaze estimation, thereby turning potential interference into reliable tracking data.
4Measurement precision
If manual camera adjustment is performed for each individual facial geometry, then measurement precision is improved, but ease of operation deteriorates
Solution Approach 1:
The patent implements self-service calibration where the system automatically adapts to each user's facial geometry without manual adjustment. The algorithm automatically detects eye features and adjusts parameters based on the captured images, enabling the system to serve itself in the calibration process. This eliminates the need for manual camera adjustment while maintaining high detection accuracy across different users.
Data Source
AI summary
A system and method for determining one or more parameters of a user's eye is disclosed. In one example, the system includes a computing unit and a head-wearable device including a first camera having a field of view, and a non-transparent component. The first camera and the non-transparent component are arranged relative to each other so that at least a portion of the eye and at least a portion of the non-transparent component are expected to be in the field of view of the first camera. The computing unit implements a neural network and is configured to receive an image from the first camera, feed the image as an input into the neural network, and determine the one or more parameters as an output of the neural network.