logo

For the Inventor, By the Inventor

The patent badge is an abbreviated version of the USPTO patent document. The patent badge does contain a link to the full patent document.

The patent badge is an abbreviated version of the USPTO patent document. The patent badge covers the following: Patent number, Date patent was issued, Date patent was filed, Title of the patent, Applicant, Inventor, Assignee, Attorney firm, Primary examiner, Assistant examiner, CPCs, and Abstract. The patent badge does contain a link to the full patent document (in Adobe Acrobat format, aka pdf). To download or print any patent click here.

Patent No.:

US 12518743 B1

PDFFull Text
Date of Patent:
Jan. 06, 2026

Filed:

Dec. 09, 2022

Computer-implemented multiscale multimodal transformer for multimodal action recognition

Applicant:

Amazon Technologies, Inc., Seattle, WA (US);

Inventors:

Wentao Zhu, Redmond, WA (US);

Keval Doshi, Seattle, WA (US);

Robert Ramsey, Seattle, WA (US);

Xiaolong Wang, Campbell, CA (US);

Mohamed Kamal Omar, Seattle, WA (US);

Ahmed Aly Saad Ahmed, Bothell, WA (US);

Assignee:

Amazon Technologies, Inc., Seattle, WA (US);

Attorney:

NICHOLSON DE VOS WEBSTER & ELLIOTT LLP

Primary Examiner:

Linda Wong

Int. Cl.
CPC ...
G10L 15/08 (2006.01); G06N 20/00 (2019.01); G06N 20/20 (2019.01); G06T 9/00 (2006.01); G10L 15/02 (2006.01); G10L 21/04 (2013.01); G10L 21/10 (2013.01);
U.S. Cl.
CPC ...
G10L 15/08 (2013.01); G06N 20/00 (2019.01); G06N 20/20 (2019.01); G06T 9/002 (2013.01); G10L 15/02 (2013.01); G10L 21/04 (2013.01); G10L 21/10 (2013.01);
Abstract

Techniques for implementing a multiscale multimodal transformer for multimodal action recognition with a computer are described. According to some examples, a computer-implemented method includes training a multiscale audio transformer (MAT) machine learning model to extract hierarchical audio representations; and generating an audio inference by the MAT machine learning model for an input audio file.

Find Patent Forward Citations

Zhu, Wentao. (2026). Computer-implemented multiscale multimodal transformer for multimodal action recognition (U.S. Patent No. 12518743). U.S. Patent and Trademark Office. https://idiyas.com/patent/badge/12518743

Data source: USPTO (United States Patent and Trademark Office) public patent records. Weekly synchronization. How IDiyas builds this profile

Loading…