Vehicle Occupant Gaze Tracking via Reflected Multi-View Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional direct camera imaging in vehicles faces challenges such as occlusions, direct sunlight, and wide or long distance head and eye positions, which impede accurate gaze tracking of multiple occupants.
Innovation Solution
Utilizing multiple cameras to capture occupant images from reflections within the vehicle, combined with deep learning models trained on reflected face and eye images, for independent gaze tracking of each occupant, and implementing image quality threshold gating to conserve processing resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If direct camera imaging is used for eye tracking in vehicles, then the system structure is simple, but accuracy deteriorates due to occlusions, direct sunlight, and wide or long distance head and eye positions
Solution Approach 1:
The patent introduces reflective surfaces (mirrors, glass surfaces) as intermediaries to capture eye region images. These reflective surfaces act as mediators between the cameras and the occupants, enabling indirect imaging that avoids direct exposure to occlusions and sunlight while maintaining tracking accuracy
Solution Approach 2:
The system transitions from direct frontal imaging to multi-angle reflected imaging by positioning cameras at different locations (dashboard, door panels, windshield) to capture reflections from various angles, adding spatial dimensionality to the imaging approach
2Measurement precision
If multiple cameras are used to capture reflected images for improved tracking, then eye tracking accuracy improves, but power consumption increases
Solution Approach 1:
The system uses multiple cameras positioned at strategic locations (dashboard, door panels, windshield) to capture reflected images from specific angles where occlusions are minimized, rather than using all possible cameras continuously, optimizing the balance between tracking accuracy and energy consumption
Solution Approach 2:
The reflective surfaces in the vehicle interior (mirrors, glass) naturally provide the imaging function without requiring additional active optical components, allowing the system to leverage existing vehicle structures for the imaging task
3Loss of information
If all camera feeds are processed for comprehensive data collection, then information completeness improves, but processing efficiency deteriorates
Solution Approach 1:
The system extracts and processes only the reflected image data from specific camera feeds that are most likely to contain usable eye region information, filtering out feeds that are less likely to provide accurate tracking data, thereby improving processing efficiency while maintaining information quality
Solution Approach 2:
Different camera feeds are evaluated and processed differently based on their specific characteristics and the quality of reflected images they capture, with more resources allocated to feeds showing high-quality eye region data and fewer resources to feeds with poor quality or occluded views
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
Improves eye tracking accuracy and efficiency by utilizing reflected images and deep learning models, reducing power consumption and enhancing gaze tracking in vehicles with reflective surfaces.
Implementation Method 1
obtaining face image data, eye region image data, and head pose data from reflected images of one or more occupants
Data Source
AI summary
Methods, systems, and storage media for performing multi-user gaze tracking in a vehicle space using multi-surface optical reflections are disclosed. Implementations may: acquire face and eye region image data of a plurality of occupants within a field of view of at least one camera associated with a vehicle; evaluate reflected image quality thresholds; locate and match occupants within the vehicle space; and perform eye tracking for multiple occupants independently via reflected multi-view images provided to a deep learning model.