US 11,748,989 B2
Enhancing detection of occluded objects in a multiple object detection system
Gopi Subramanian, Delray Beach, FL (US); and Joseph Celi, Boca Raton, FL (US)
Assigned to Sensormatic Electronics, LLC, Boca Raton, FL (US)
Filed by Sensormatic Electronics, LLC, Boca Raton, FL (US)
Filed on Aug. 26, 2022, as Appl. No. 17/822,647.
Application 17/822,647 is a continuation of application No. 17/019,038, filed on Sep. 11, 2020, granted, now 11,430,221.
Prior Publication US 2022/0415049 A1, Dec. 29, 2022
This patent is subject to a terminal disclaimer.
Int. Cl. G06K 9/00 (2022.01); G06V 20/40 (2022.01); G06T 7/246 (2017.01); G06V 20/58 (2022.01); G06V 40/16 (2022.01); G06T 5/40 (2006.01)
CPC G06V 20/46 (2022.01) [G06T 5/40 (2013.01); G06T 7/246 (2017.01); G06V 20/58 (2022.01); G06V 40/161 (2022.01); G06T 2207/10016 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
detecting, via an object detection system, a plurality of current objects in a current video frame;
generating initial object identifier mapping information, the initial object identifier mapping information assigning a candidate object identifier to a first object of the plurality of current objects;
generating, based a multi-object tracking (MOT) inaccuracy condition, a current fingerprint corresponding to the first object of the plurality of current objects;
determining that the current fingerprint corresponds to a historic fingerprint associated with a first object of a plurality of historic objects detected in a previous video frame, the first object of the plurality of historic objects associated with a final object identifier;
generating updated object identifier mapping information, the updated object identifier mapping information assigning the final object identifier to the first object of the plurality of current objects; and
generating tracking information based on the updated object identifier mapping information, the tracking information corresponding to a trajectory of the first object between the previous video frame and the current video frame for the first object.