	US 12,131,497 B2
	Fast object search based on the cocktail party effect
Fangyi Chen, Pittsburgh, PA (US); Shayeree Sarkar, Pittsburgh, PA (US); and Marios Savvides, Pittsburgh, PA (US)
Assigned to Carnegie Mellon University, Pittsburgh, PA (US)
Appl. No. 18/266,737
Filed by CARNEGIE MELLON UNIVERSITY, Pittsburgh, PA (US)
PCT Filed Jan. 31, 2022, PCT No. PCT/US2022/014558 § 371(c)(1), (2) Date Jun. 12, 2023, PCT Pub. No. WO2022/173607, PCT Pub. Date Aug. 18, 2022.
Claims priority of provisional application 63/147,449, filed on Feb. 9, 2021.
Prior Publication US 2024/0046503 A1, Feb. 8, 2024
Int. Cl. G06V 10/774 (2022.01); G06T 7/70 (2017.01)

CPC G06T 7/70 (2017.01) [G06V 10/774 (2022.01); G06V 2201/07 (2022.01)]

10 Claims

1. A method comprising:

grouping a plurality of images into multiple grouped images;

inputting each grouped image to an object detection model trained on grouped images;

determining that a grouped image contains an object-of-interest for which the object detection model has been trained;

determining the images comprising the grouped image; and

inputting each of the images to an object detector trained to detect the objects-of-interest in images;

wherein grouping a plurality of images into multiple grouped images comprises:

selecting n images; and

performing a channel-wise concatenation of the n images.