Eye tracking method and system

Inventors

De Villers-Sidani, Etienne • DROUIN-PICARO, Paul Alexandre

Assignees

Innodem Neurosciences

Abstract

A computer-implemented method for determining a gaze position of a user, comprising: receiving an initial image of at least one eye of the user; extracting at least one color component of the initial image to obtain a corresponding at least one component image; for each component image, determining a respective internal representation; determining an estimated gaze position in the initial image by applying a respective primary stream to obtain a respective internal representation for each of the at least one component image; and outputting the estimated gaze position. The processing of the component images is performed using a neural network configured to, at run time and after the neural network has been trained, process the component images using one or more neural network layers to generate the estimated gaze position. A system for determining a gaze position of a user is also provided.

Core Innovation

The invention relates to determining a gaze position of a user from an initial image of at least one eye. From the initial eye image, at least one color component is extracted to obtain a corresponding at least one component image. A respective primary stream is applied to each component image to obtain a respective internal representation for each component image.

The internal representations are fused by an internal stream having fusion layers, with optional auxiliary inputs including estimated illuminant values and facial landmark coordinates or head-pose features. The model outputs an estimated gaze position in image or screen coordinates, with optional separate determination of the first and second gaze coordinates.

The disclosure further includes calibration using calibration images associated with a calibration position. Selectable calibration models are described, including a calibration neural network, ridge regression, decision trees, a support vector machine, and linear regression, together with image orientation determination relative to a reference and orientation-specific prediction streams or modules for orientation-aware gaze position estimation.

Claims Coverage

The document includes three independent claims that cover a computer-implemented method, a system, and another computer-implemented method. Across these claims, the core coverage includes extraction of color components from an initial eye image, primary-stream processing to internal representations, and gaze position estimation, with dependent-claim refinements addressing fusion, coordinate-wise estimation, calibration, regression-based determination, and orientation modules.

Overall, the independent claims are centered on color-component extraction from an initial eye image, primary-stream processing to obtain internal representations or per-component gaze positions, and gaze position estimation for the initial image based on those representations or positions. The claim set also supports refinements including fusion, coordinate-wise handling, calibration with selectable calibration models, and orientation module processing.

Stated Advantages

Documented Applications

No documented applications found