US 11,934,955 B2
Neural network based facial analysis using facial landmarks and associated confidence values
Nuri Murat Arar, Zurich (CH); Niranjan Avadhanam, Saratoga, CA (US); Nishant Puri, San Francisco, CA (US); Shagan Sah, Santa Clara, CA (US); Rajath Shetty, Santa Clara, CA (US); Sujay Yadawadkar, Santa Clara, CA (US); and Pavlo Molchanov, Mountain View, CA (US)
Assigned to NVIDIA Corporation, Santa Clara, CA (US)
Filed by NVIDIA Corporation, Santa Clara, CA (US)
Filed on Oct. 31, 2022, as Appl. No. 18/051,296.
Application 18/051,296 is a continuation of application No. 17/004,252, filed on Aug. 27, 2020, granted, now 11,487,968.
Claims priority of provisional application 62/948,796, filed on Dec. 16, 2019.
Prior Publication US 2023/0078171 A1, Mar. 16, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 3/08 (2023.01); G06F 18/21 (2023.01); G06F 18/214 (2023.01); G06N 20/00 (2019.01); G06V 10/764 (2022.01); G06V 10/774 (2022.01); G06V 10/82 (2022.01); G06V 10/94 (2022.01); G06V 20/59 (2022.01); G06V 20/64 (2022.01); G06V 40/16 (2022.01); G06V 40/18 (2022.01)
CPC G06N 3/08 (2013.01) [G06F 18/214 (2023.01); G06F 18/2193 (2023.01); G06N 20/00 (2019.01); G06V 10/764 (2022.01); G06V 10/774 (2022.01); G06V 10/82 (2022.01); G06V 10/95 (2022.01); G06V 20/597 (2022.01); G06V 20/647 (2022.01); G06V 40/171 (2022.01); G06V 40/193 (2022.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
determining, using one or more machine learning models (MLMs), and based at least on first image data representing a first portion of an image, one or more landmark points for a subject at least partially depicted in the first portion;
determining one or more confidence values corresponding to amounts of occlusion depicted in the first portion of the image corresponding to the one or more landmark points;
selecting a second portion of the image based at least on one or more portions of the subject in the image detected using the one or more confidence values;
computing one or more properties of the subject based at least on analyzing second image data representing the second portion of the image; and
initiating one or more operations of a machine based at least on the one or more properties of the subject.