US 12,464,199 B2
System and method for automated video generation for training of machine learning algorithms in aviation environments
Bikramjit Karmakar, Bangalore (IN); Rajdeep Sawant, Hyderabad (IN); Yogesh Selvaraj, Bangalore (IN); Amrit Chatterjee, Hyderabad (IN); Lennon Ignatius Dhanaraj, Bangalore (IN); Hemant Kumar, Hyderabad (IN); Victor Vishal Rana, Bangalore (IN); and Subhashish Chakravarty, Marion, IA (US)
Assigned to B/E Aerospace, Inc., Winston-Salem, NC (US)
Filed by B/E Aerospace, Inc., Winston-Salem, NC (US)
Filed on Mar. 2, 2023, as Appl. No. 18/116,706.
Claims priority of application No. 202241041633 (IN), filed on Jul. 20, 2022.
Prior Publication US 2024/0031658 A1, Jan. 25, 2024
Int. Cl. H04N 21/81 (2011.01); G06T 7/50 (2017.01); G06T 7/70 (2017.01); G06T 13/40 (2011.01); G06V 20/59 (2022.01)
CPC H04N 21/816 (2013.01) [G06T 7/50 (2017.01); G06T 7/70 (2017.01); G06T 13/40 (2013.01); G06V 20/59 (2022.01)] 18 Claims
OG exemplary drawing
 
1. A computer-implemented method for automated video generation, the method comprising:
receiving, via an automated video generation system, a desired scenario comprising one or more character actions to be simulated within an aviation environment;
generating, via a graphics engine, a plurality of digital human characters configured for execution of the one or more character actions, each digital human character having a photorealistic human appearance, one or more human attributes, and at least one aviation attribute;
generating, via the graphics engine, a photorealistic representation of the aviation environment comprising:
an interior or exterior aviation space;
one or more fixtures within the interior or exterior aviation space;
one or more free objects disposed within the interior or exterior aviation space;
and
one or more environmental attributes including at least one of a lighting configuration, a physics configuration, and a camera view;
generating, via the graphics engine, one or more character movements based on the one or more character actions, each character movement comprising an animated sequence of character poses;
and
generating, via a video automation pipeline, at least one video dataset based on the desired scenario, the at least one video dataset animating the one or more character actions performed by the plurality of digital human characters within the aviation environment and including at least one annotation file comprising:
pixel location data corresponding to each digital human character, each fixture, and each free object;
depth data corresponding to each digital human character, each fixture, and each free object;
and
annotation data corresponding to each character action.