GPU Eye Tracking Offloading CPU Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current eye tracking systems for computers are computationally complex and resource-intensive, causing them to interfere with other processes and slow down system performance, making them unsuitable for integration into standard personal computers without compromising other functions.
Innovation Solution
Offloading data processing tasks from the CPU to the graphics card, utilizing it as a parallel processor to determine the gaze point by generating and processing screen patterns reflected on the eye, thereby reducing CPU load and optimizing resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If eye tracking is implemented using standard CPU processing, then gaze point detection can be achieved, but system performance degrades and other processes are slowed down
Solution Approach 1:
The patent introduces a graphics card as an intermediary processing unit between the camera and the CPU. The graphics card receives image data from the camera, performs eye tracking processing using its parallel processing capabilities, and outputs results to the CPU. This mediator handles the computationally intensive tasks, preventing system performance degradation while maintaining accurate gaze point detection.
Solution Approach 2:
The patent replaces the traditional CPU-based processing mechanism with a graphics card-based parallel processing mechanism. By utilizing the graphics card's GPU architecture designed for parallel computations, the system efficiently handles the complex image processing and eye tracking algorithms without burdening the main CPU, thus resolving the contradiction between detection accuracy and system performance.
2Speed
If eye tracking processing is performed in real-time, then responsive gaze tracking is achieved, but computational complexity increases
Solution Approach 1:
The patent substitutes the sequential processing architecture of traditional CPUs with the parallel processing architecture of graphics cards. This architectural change enables real-time processing of eye tracking data by simultaneously processing multiple image pixels and features, achieving both real-time speed and reduced computational complexity through efficient parallel computation.
Solution Approach 2:
The patent transitions from single-threaded sequential processing to multi-threaded parallel processing by utilizing the graphics card's parallel architecture. This dimensional change in processing approach allows multiple computational tasks to execute simultaneously, achieving real-time performance without exponentially increasing overall computational complexity.
3Measurement precision
If CPU resources are dedicated to eye tracking, then accurate gaze detection is achieved, but other computer functions are impaired
Solution Approach 1:
The graphics card serves as an intermediary that handles all eye tracking processing tasks independently from the CPU. This separation ensures that the CPU remains available for other computer functions while the graphics card dedicately processes eye tracking data with sufficient resources to maintain accurate gaze detection without impairing overall system versatility.
Solution Approach 2:
The patent segments the computer's processing responsibilities by assigning eye tracking tasks to the graphics card while leaving the CPU free for other functions. This segmentation of processing duties allows both eye tracking accuracy and overall system functionality to be maintained simultaneously through specialized processing units.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided is a method of determining a gaze point of an eye watching a visual display controllable by a display signal. The method comprises generating a display signal using a graphics card in order for the visual display to produce a screen pattern; receiving a signal encoding an image of the eye including a corneo-scleral reflection of the screen pattern; and determining, based on in part the geometry of said reflection, a gaze point of the eye, wherein said determining a gaze point includes utilising the graphics card as a parallel processor. The image of the eye may be received directly at the graphics card. The graphics card may extract image features in the eye images. Reference illuminators may be used, and the screen pattern may be interlaced with a distinctive reference pattern. Further provided are a gaze-tracking system and a personal computer system adapted to determine a gaze point of a viewer.