IP Library › Granted Patent US 11,557,134
Granted Patent B2
US 11,557,134 · App. 17/132,017 · Granted Jan 17, 2023

Methods and systems for training an object detection algorithm using synthetic images

Inventors: Ivo Moravec (Richmond Hill, CA); Jie Wang (Markham, CA); Syed Alimul Huda (Scarborough, CA)
Assignee: SEIKO EPSON CORPORATION
G06V20/64G06K9/6262G06T7/70G06T19/006G06V20/20G06T2207/20081
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 11,557,134
App. No.
17/132,017
Granted
Jan 17, 2023
Kind
B2
Abstract

A method includes: (A) receiving a selection of a 3D model stored in one or more memories, the 3D model corresponding to an object and (B) setting a camera parameter set for a camera for use in detecting a pose of the object in a real scene. The method also includes (C) generating at least one 2D synthetic image based at least on the camera parameter set by rendering the 3D model in a view range for generating training data.

Claims (28)

1. A method for training an object detection algorithm, comprising:

(A) receiving a selection of a 3D model stored in one or more memories, the 3D model corresponding to an object;

(B) setting a camera parameter set for a camera for use in detecting a pose of the object in a real scene by:

receiving information identifying an object detection device including the camera,

acquiring, based at least in part on the information identifying the object detection device, the camera parameter set for the object detection device from a plurality of the camera parameter sets stored in one or more memories, wherein each camera parameter set of the plurality of camera parameter sets is associated in the one or more memories with at least one object detection device of a plurality of different object detection devices; and

(C) generating at least one 2D synthetic image based at least on the camera parameter set by rendering the 3D model in a view range for generating training data.

2. The method according to claim 1 , wherein the object detection device is a head-mounted display device including the camera.

3. The method according to claim 1 , wherein:

(B) includes: setting a plurality of camera parameter sets for a plurality of cameras for use in detecting a pose of the object in a real scene; and

(C) includes: generating a plurality of 2D synthetic images based at least on the respective camera parameter sets by rendering the 3D model in the view range.

4. The method according to claim 1 , further comprising receiving, before (C), a selection of data representing the view range.

5. The method according to claim 1 , wherein the view range comprises a predetermined view range covering equal to or less than 360 degrees of azimuth and elevation.

6. The method according to claim 1 , further comprising:

(D) receiving, via a graphical user interface, a selection of data representing the view range.

7. The method according to claim 6 , wherein (D) comprises receiving the selection of data representing the view range from a user selection on a graphical user interface including a preview view of the object and a graphical representation of the user selected view range.

8. A method for training an object detection algorithm, comprising:

(A) receiving a selection of a 3D model stored in one or more memories, the 3D model corresponding to an object;

(B) setting a camera parameter set for a camera for use in detecting a pose of the object in a real scene by: acquiring, through a data connection, the camera parameter set from an object detection device having the camera when the object detection device becomes accessible by the one or more processors through the data connection; and

(C) generating at least one 2D synthetic image based at least on the camera parameter set by rendering the 3D model in a view range for generating training data.

9. The method according to claim 8 , wherein the object detection device is a head-mounted display device including the camera.

10. The method according to claim 8 , wherein:

(B) includes: setting a plurality of camera parameter sets for a plurality of cameras for use in detecting a pose of the object in a real scene; and

(C) includes: generating a plurality of 2D synthetic images based at least on the respective camera parameter sets by rendering the 3D model in the view range.

11. The method according to claim 8 , further comprising receiving, before (C), a selection of data representing the view range.

12. The method according to claim 8 , wherein the view range comprises a predetermined view range covering equal to or less than 360 degrees of azimuth and elevation.

13. The method according to claim 8 , further comprising:

(D) receiving, via a graphical user interface, a selection of data representing the view range.

14. The method according to claim 12 , wherein (D) comprises receiving the selection of data representing the view range from a user selection on a graphical user interface including a preview view of the object and a graphical representation of the user selected view range.

Continuity (3)
Continuation 16572750 · Sep 17, 2019
Continuation 15839247 · Dec 12, 2017
Related Publication 20210110141A1 · Apr 15, 2021