US 12,114,075 B1
Object selection in computer vision
Yu Shu, Foster City, CA (US); Yahia Chafik Chehadeh, Palo Alto, CA (US); Heliben Dhrumil Desai, San Jose, CA (US); and Young Sun Kim, Dublin, CA (US)
Assigned to AMAZON TECHNOLOGIES, INC., Seattle, WA (US)
Filed by Amazon Technologies, Inc., Seattle, WA (US)
Filed on Jul. 5, 2023, as Appl. No. 18/218,199.
Application 18/218,199 is a continuation of application No. 17/089,322, filed on Nov. 4, 2020, granted, now 11,743,588.
Claims priority of provisional application 63/081,422, filed on Sep. 22, 2020.
Int. Cl. H04N 23/695 (2023.01); G06V 40/10 (2022.01); H04N 7/14 (2006.01); G10L 15/22 (2006.01)
CPC H04N 23/695 (2023.01) [G06V 40/10 (2022.01); H04N 7/141 (2013.01); G10L 15/22 (2013.01); G10L 2015/223 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
detecting a first person represented in a first video data captured by a camera of a first device;
determining a first level of engagement of the first person with the camera at a first time;
determining a first ratio proportional to the first level of engagement of the first person with the camera;
determining a first weight value for the first person at a second time based at least in part by increasing an initial weight value for the first person at a first rate, wherein the first rate is determined based at least in part on the first ratio; and
controlling the first device to track the first person within frames of image data generated by the first device based at least in part on a threshold weight value.