US 12,190,574 B2
Object location determination
Alfredo Fanghella, Stockholm (SE); and Manish Sonal, Sollentuna (SE)
Assigned to TELEFONAKTIEBOLAGET LM ERICSSON (PUBL), Stockholm (SE)
Appl. No. 17/428,863
Filed by Telefonaktiebolaget LM Ericsson (publ), Stockholm (SE)
PCT Filed Feb. 21, 2019, PCT No. PCT/EP2019/054349
§ 371(c)(1), (2) Date Aug. 5, 2021,
PCT Pub. No. WO2020/169199, PCT Pub. Date Aug. 27, 2020.
Prior Publication US 2022/0157039 A1, May 19, 2022
Int. Cl. G06T 7/11 (2017.01); G06T 7/70 (2017.01); G06V 10/10 (2022.01); G06V 10/44 (2022.01); G06V 10/52 (2022.01); G06V 10/77 (2022.01); G06V 10/98 (2022.01); G06V 20/20 (2022.01)
CPC G06V 10/98 (2022.01) [G06T 7/11 (2017.01); G06T 7/70 (2017.01); G06V 10/17 (2022.01); G06V 10/44 (2022.01); G06V 10/52 (2022.01); G06V 10/7715 (2022.01); G06V 20/20 (2022.01)] 20 Claims
OG exemplary drawing
 
1. An object locator, comprising:
processing circuitry; and
memory containing instructions executable by the processing circuitry whereby the processing circuitry is operative to:
apply at least one part-based object detector to a picture to detect object parts in the picture and generate, for each detected object part, a part location representation defining a region of the picture;
estimate a size of an object comprising the object parts in the picture based on a geometric model of the object and of the object parts and based on the part location representations, wherein the geometric model defines a geometry of the object and its included object parts;
determine, based on the part location representations, search locations in the picture for a search window having a size that is based on the estimated size of the object; and
for at least a portion of the determined search locations:
identify any detected object part within the search window positioned at the determined search location;
estimate a homography mapping object part(s) in the geometric model to the identified detected object part(s) by minimizing an error between the mapped object part(s) and the identified detected object part(s); and
determine, when the error is smaller than a threshold value and for the object, an object location representation defining a region of the picture based on the homography and the geometric model.