Patent · US Active

Deep neural network pose estimation system

US11893789B2 · kind B2 · utility

2Cited by

8References

20Claims

0Family size

Assignee

Magic Leap, Inc. · US

Inventors

Danying Hu · Sunnyvale, US
Daniel DeTone · San Francisco, US
Tomasz Jan Malisiewicz · Mountain View, US

Key dates

Filing date	Nov 14, 2019
Grant date	Feb 6, 2024
Priority date	—
Expiry date	Nov 28, 2040

Classification

Technology area (CPC G)Physics
CPC primaryG06T2207/30244
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

A deep neural network provides real-time pose estimation by combining two custom deep neural networks, a location classifier and an ID classifier, with a pose estimation algorithm to achieve a 6D0F location of a fiducial marker. The locations may be further refined into subpixel coordinates using another deep neural network. The networks may be trained using a combination of auto-labeled videos of the target marker, synthetic subpixel corner data, and/or extreme data augmentation. The deep neural network provides improved pose estimations particularly in challenging low-light, high-motion, and/or high-blur scenarios.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.