Granted Patent B2

US 12,554,319 · App. 18/424,924 · Granted Feb 17, 2026

Inventors: Anders Dahl (Danderyd, SE); Erik Lindén (Danderyd, SE); Anna Larsen Redz (Danderyd, SE); Per Olof Jonatan Walck (Danderyd, SE); Pontus Christian Walck (Danderyd, SE)

G06F3/012G06T7/70G06V10/762G06V20/70G06T2207/30196G06T2207/30204

Abstract

Claims (53)

1 . A system for tracking a pose of a head of a user and generating training data, the system comprising a motion-capture head module for tracking a head pose of a user, the motion-capture head module comprising:

a flexible headband for attaching to a head of the user;

a post extending from the headband;

at least three markers, wherein each marker is coupled to the post by a respective one of at least three branching elements extending from the post, such that the at least three markers have a first fixed geometrical relationship with respect to each other; wherein the system further comprises:

at least one camera configured to capture images of the user wearing the motion-capture head module from a known pose in a world coordinate system;

a processing unit configured to:

triangulate three-dimensional (3D) coordinates of the markers based on images captured from at least two cameras at a given time;

determine a deformation of the flexible headband based on the triangulated 3D coordinates of one or more markers;

determine a 3D profile of the headband in a head module coordinate system based on the deformation of the headband;

determine a pose of the head module coordinate system with respect to the world coordinate system;

determine a two-dimensional (2D) profile of the headband in an image based on the pose of the head module coordinate system and the pose of the respective camera in the world coordinate system; and

generate a training image by painting over the 2D profile of the headband in the image.

2 . The system of claim 1 , wherein the post comprises a rigid post and the at least three branching members comprise rigid branching members.

3 . The system of claim 1 , wherein the at least three markers comprise at least three reflective markers.

4 . The system of claim 1 , wherein the first fixed geometrical relationship comprises an irregular relationship.

5 . The system of claim 1 , wherein each marker comprises a different size.

6 . The system of claim 1 , wherein the motion-capture head module further comprises:

a second post extending from the headband; and

at least three further markers, wherein each further marker is coupled to the second post by a respective one of at least three further branching members extending from the second post, such that the at least three further markers have a second fixed geometrical relationship with respect to each other.

7 . The system of claim 6 , wherein the second fixed geometrical relationship is different to the first fixed geometrical relationship.

8 . A method for tracking a pose of a head of a user, the method comprising:

receiving a first image, captured by a first camera at a first time, of the user wearing a motion-capture head module, wherein the motion capture head module comprises:

a headband;

a first cluster of markers comprising a first marker, a second marker and a third marker having a first fixed geometrical relationship with respect to each other; and

a second cluster of markers comprising a fourth marker, a fifth marker and a sixth marker, having a second fixed geometrical relationship with respect to each other;

receiving a second image, captured by a second camera at the first time, of the user wearing the motion-capture head module, wherein the first camera and the second camera each have a different known pose in a world coordinate system and wherein the first to sixth markers are visible in both the first and second images;

triangulating three-dimensional (3D) coordinates of each marker in each of the first cluster and the second cluster in the world coordinate system, based on the position of each marker in each of the first image and the second image;

individually associating each triangulated 3D coordinate with a respective one of the first to sixth markers using the first and second fixed geometrical relationships;

determining a third fixed geometrical relationship between each of the first to sixth markers based on the specifically associated 3D coordinates of each marker;

receiving further first images captured by the first camera at a series of time points and receiving corresponding further second images captured by the second camera at the series of time points, wherein each further first image and each further second image capture the user wearing the motion-capture head module;

for each corresponding pair of further first and second images:

triangulating the 3D coordinates of at least three markers of the first to sixth markers in the world coordinate system based on the position of each of the at least three markers in the further first image and the further second image; and

determining a pose of a head module coordinate system with respect to the world coordinate system based on the 3D coordinates of the at least three markers and the third fixed geometrical relationship;

wherein the headband is a flexible headband, and the method further comprises:

determining a deformation of the headband based on the individually associated 3D coordinates of one or more markers in each of the first cluster and the second cluster;

determining a 3D profile of the headband in the head module coordinate system based on the deformation of the headband;

determining a two-dimensional, 2D, profile of the headband in each of the further first images and further second images based on:

a pose of the head module coordinate system with respect to the world coordinate system in the respective further first image or further second image;

a pose of the respective first or second camera in the world coordinate system; and

generating first and second training images by painting over the 2D profile of the headband in each of the respective further first images and further second images.

9 . The method of claim 8 , wherein the second fixed geometrical relationship is different to the first fixed geometrical relationship.

10 . The method of claim 8 , wherein each marker comprises a different size.

11 . The method of claim 8 , wherein the first fixed geometrical relationship and the second fixed geometrical relationship comprise an irregular relationship.

12 . A method of generating training images for a machine learning head-pose detection algorithm, the method comprising:

performing the method of tracking a pose of a head of a user according to claim 9 ;

determining a head pose value for each further first image and further second image based on the respective pose of the head module coordinate system and the pose of the first camera and the second camera in the world coordinate system;

determining a deformation of the flexible headband based on the individually associated 3D coordinates of one or more markers in each of the first cluster and the second cluster;

determining a 3D profile of the headband in the head module coordinate system based on the deformation of the headband;

determining a two-dimensional, 2D, profile of the headband in each of the further first images and further second images based on:

a pose of the head module coordinate system with respect to the world coordinate system in the respective further first image or further second image;

a pose of the respective first or second camera in the world coordinate system; and

generating first and second training images by painting over the 2D profile of the headband in each of the respective further first images and further second images and annotating them with the respective head pose value.

13 . The method of claim 12 , wherein determining the head pose value for each further first image and further second image based on the respective pose of the head module coordinate system comprises determining a pose of a head coordinate system with respect to the world coordinate system by transforming the head module coordinate system using a predetermined mapping operator.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (25)