US 11,656,748 B2
Machine learning in video classification with playback highlighting
Reza Zadeh, Emerald Hills, CA (US); Dong Wang, San Jose, CA (US); Deepak Menghani, Mountain View, CA (US); John Goddard, Palo Alto, CA (US); and Ryan Tobin, Sacramento, CA (US)
Assigned to MATROID, INC., Palo Alto, CA (US)
Filed by Matroid, Inc., Palo Alto, CA (US)
Filed on Dec. 10, 2021, as Appl. No. 17/547,259.
Application 17/547,259 is a continuation of application No. 16/945,805, filed on Aug. 1, 2020, granted, now 11,232,309.
Application 16/945,805 is a continuation of application No. 15/905,679, filed on Feb. 26, 2018, granted, now 10,789,291, issued on Sep. 29, 2020.
Claims priority of provisional application 62/465,788, filed on Mar. 1, 2017.
Claims priority of provisional application 62/465,790, filed on Mar. 1, 2017.
Prior Publication US 2022/0101008 A1, Mar. 31, 2022
Int. Cl. G06F 3/048 (2013.01); G06F 3/04842 (2022.01); G11B 27/34 (2006.01); G11B 27/36 (2006.01); G06F 16/74 (2019.01); G06F 16/783 (2019.01); G06F 16/2457 (2019.01); G06F 3/00 (2006.01); G06F 16/738 (2019.01); G06V 20/40 (2022.01); G06F 18/241 (2023.01); G06V 10/764 (2022.01); G06F 3/0484 (2022.01)
CPC G06F 3/04842 (2013.01) [G06F 3/00 (2013.01); G06F 16/24578 (2019.01); G06F 16/738 (2019.01); G06F 16/74 (2019.01); G06F 16/7837 (2019.01); G06F 18/241 (2023.01); G06V 10/764 (2022.01); G06V 20/41 (2022.01); G11B 27/34 (2013.01); G11B 27/36 (2013.01); G06F 3/0484 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
receiving, from a user, a request to identify a selected object in a selected video;
identifying a set of frames of the selected video that include the selected object using a classifier trained to identify frames that have an above-threshold confidence of including the selected object;
modifying an interface in which the selected video is played by highlighting the selected object within the frame with an overlaid interface element that moves within the displayed video to follow the selected object as the selected object moves within the video;
providing for display to the user a feedback interface element;
receiving, via the feedback interface element, an input from the user indicating either that a displayed frame with an overlaid interface element does not include the selected object or that a displayed frame without an overlaid interface element does include the selected object; and
retraining the classifier based on the received input.