IP Library › Granted Patent US 12,416,972
Granted Patent B2
US 12,416,972 · App. 18/757,384 · Granted Sep 16, 2025

Method of performing a user-specific and device-specific calibration of point of gaze estimation

Inventors: Karol Duzinkiewicz (Banino, PL); Jan Glinko (Gdansk, PL); Artur Skrzynecki (Wejherowo, PL); Michael Schiessl (Berlin, DE); Cezary Polak (Gdynia, PL)
Assignee: eye square GMBH
G06F3/013G06T7/80
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,416,972
App. No.
18/757,384
Granted
Sep 16, 2025
Kind
B2
Abstract

A method of performing a user-specific and device-specific calibration of point of gaze estimation comprising a user's mobile device, a calibration target displayer, a built-in camera video data recorder, a perspective transform matrix process, a calibration data set splitter, and a support vector regression calculator by having a built-in camera process gaze in absolute measurement terms based on a series of successive data points based on a non-static marker.

Claims (51)

1. A method of calibrating point of gaze estimation, comprising the steps:

providing a calibration application;

providing a user device;

wherein said user device comprises a display and a user-facing camera;

wherein said calibration application runs on said user device;

wherein said calibration application causes to be displayed on said display a non-static calibration marker;

prompting a user to look at said non-static calibration marker by displaying said non-static calibration marker on said display;

capturing one or more images of said user as said user is prompted to look at said non-static calibration marker;

recording a series of coordinates corresponding to where said non-static calibration marker appears on said display when said one or more images of said user are captured;

wherein said series of coordinates comprise a series of target ground truth positions for said non-static calibration marker, which comprise a target ground truth position data;

generating a calibration data set by matching said target ground truth position data with said captured images; and

processing said calibration data set by a perspective transform matrix process into a 3×3 transform matrix.

2. The method of calibrating point of gaze estimation of claim 1 , wherein said target ground truth position data comprises one or more vertices.

3. The method of calibrating point of gaze estimation of claim 2 , wherein said one or more vertices comprise four vertices.

4. The method of calibrating point of gaze estimation of claim 3 , wherein said four vertices comprise a first vertex, a second vertex, a third vertex, and a fourth vertex; and

wherein said one or more vertices, if connected, form a quadrangle.

5. The method of calibrating point of gaze estimation of claim 4 , wherein said first vertex is a top left display corner;

wherein said second vertex is a top right display corner;

wherein said third vertex is a bottom left display corner; and

wherein said fourth vertex is a bottom right display corner.

6. The method of calibrating point of gaze estimation of claim 5 , wherein where said user is determined to have been looking is a look point;

dividing an input cloud into four quadrants, wherein said input cloud comprises said look points.

7. The method of calibrating point of gaze estimation of claim 6 , wherein a center point of said four quadrants comprises a center of mass of said look points.

8. The method of calibrating point of gaze estimation of claim 7 , wherein said vertices are determined to produce a quadrangle of the largest area for a given quadrant.

9. The method of calibrating point of gaze estimation of claim 8 , wherein said user comprises one or more user characteristics;

wherein said one or more user characteristics comprise gender, skin color, eye shape, age, hair, background color, and lighting conditions; and

wherein said user characteristics are used as factors in said perspective transform matrix.

10. A method of calibrating point of gaze estimation, comprising the steps:

providing a calibration application;

providing a user device;

wherein said user device comprises a display and a user-facing camera;

wherein said calibration application runs on said user device;

wherein said calibration application causes to be displayed on said display a non-static calibration marker;

prompting a user to look at said non-static calibration marker by displaying said non-static calibration marker on said display;

capturing one or more images of said user as said user is prompted to look at said non-static calibration marker;

recording a series of coordinates corresponding to where said non-static calibration marker appears on said display when said one or more images of said user are captured;

wherein said series of coordinates comprise a series of target ground truth positions for said non-static calibration marker, which comprise a target ground truth position data;

generating a calibration data set by matching said target ground truth position data with said captured images;

processing said calibration data set by a perspective transform matrix process into a 3×3 transform matrix;

wherein said target ground truth position data comprises one or more vertices;

wherein said one or more vertices comprise four vertices;

wherein said four vertices comprise a first vertex, a second vertex, a third vertex, and a fourth vertex;

wherein said one or more vertices, if connected, form a quadrangle;

wherein said first vertex is a top left display corner;

wherein said second vertex is a top right display corner;

wherein said third vertex is a bottom left display corner;

wherein said fourth vertex is a bottom right display corner;

wherein where said user is determined to have been looking is a look point;

dividing an input cloud into four quadrants, wherein said input cloud comprises said look points;

wherein a center point of said four quadrants comprises a center of mass of said look points;

wherein said vertices are determined to produce a quadrangle of the largest area for a given quadrant.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jul 9, 2025
From: DUZINKIEWICZ, KAROL; GLINKO, JAN; SKRZYNECKI, ARTUR; SCHIESSL, MICHAEL; POLAK, CEZARY
To: EYE SQUARE, A GERMAN LIMITED LIABILITY COMPANY
Reel/Frame 071645/0324 →
Continuity (2)
Continuation In Part 18345911 · Jun 30, 2023
Related Publication 20250004548A1 · Jan 2, 2025
References Cited (6)
US 10884494B1 · Lagies · 2021 [cited by examiner]
US 20180197336A1 · Rochford · 2018 [cited by examiner]
US 20220050521A1 · Drozdov · 2022 [cited by examiner]
US 20220206571A1 · Drozdov · 2022 [cited by examiner]
US 20220236797A1 · Drozdov · 2022 [cited by examiner]
US 20240094808A1 · Fu · 2024 [cited by examiner]