US 12,469,263 B2
Learning system, object detection system, learning method, and computer program
Takahiro Toizumi, Tokyo (JP)
Assigned to NEC CORPORATION, Tokyo (JP)
Appl. No. 18/016,579
Filed by NEC Corporation, Tokyo (JP)
PCT Filed Jul. 27, 2020, PCT No. PCT/JP2020/028682
§ 371(c)(1), (2) Date Jan. 17, 2023,
PCT Pub. No. WO2022/024178, PCT Pub. Date Feb. 3, 2022.
Prior Publication US 2023/0316732 A1, Oct. 5, 2023
Int. Cl. G06V 10/776 (2022.01); G06T 7/11 (2017.01); G06V 10/44 (2022.01)
CPC G06V 10/776 (2022.01) [G06T 7/11 (2017.01); G06V 10/443 (2022.01); G06V 2201/02 (2022.01)] 8 Claims
OG exemplary drawing
 
1. A learning system comprising:
at least one memory that is configured to store instructions; and
at least one first processor that is configured to execute the instructions to;
generate an area candidate that is a candidate for an area in which an object is detected from an image;
obtain information about a ground truth area including the object; and
generate a ground truth target used for learning of an object detector that detects the object from the image, on the basis of a score indicating a degree of area matching between the area candidate and the ground truth area and a size of the ground truth area,
wherein the at least one first processor that is configured to execute the instructions to:
change a score threshold, which is set for the score, in accordance with the size of the ground truth area, and
generate the ground truth target based on a result of comparison between the score and the score threshold,
wherein a range of the score threshold is higher based on the size of the ground truth area being larger, and the range of the score threshold is lower based on the size of the ground truth area being smaller, and
wherein the ground truth target is a binary bitmap.