US 11,809,525 B2
List generation device, photographic subject identification device, list generation method, and program
Takashi Hosono, Tokyo (JP); Yukito Watanabe, Tokyo (JP); Jun Shimamura, Tokyo (JP); and Atsushi Sagata, Tokyo (JP)
Assigned to NIPPON TELEGRAPH AND TELEPHONE CORPORATION, Tokyo (JP)
Appl. No. 17/312,367
Filed by NIPPON TELEGRAPH AND TELEPHONE CORPORATION, Tokyo (JP)
PCT Filed Dec. 2, 2019, PCT No. PCT/JP2019/047005
§ 371(c)(1), (2) Date Jun. 9, 2021,
PCT Pub. No. WO2020/121866, PCT Pub. Date Jun. 18, 2020.
Claims priority of application No. 2018-231867 (JP), filed on Dec. 11, 2018.
Prior Publication US 2022/0019841 A1, Jan. 20, 2022
Int. Cl. G06F 18/22 (2023.01); G06F 16/55 (2019.01); G06F 16/532 (2019.01); G06V 10/98 (2022.01); G06T 7/60 (2017.01); G06F 18/231 (2023.01); G06F 18/241 (2023.01)
CPC G06F 18/22 (2023.01) [G06F 16/532 (2019.01); G06F 16/55 (2019.01); G06F 18/231 (2023.01); G06F 18/241 (2023.01); G06T 7/60 (2013.01); G06V 10/98 (2022.01)] 14 Claims
OG exemplary drawing
 
1. A list generation apparatus comprising a processor configured to execute a method comprising:
extracting a first feature from a first image of a plurality images, wherein the first image indicates a first object, and wherein the first feature is associated with a first attribute of the first object;
extracting a second feature from a second image of the plurality of images, wherein the second image indicates a second object, the second object is distinct from the first object, and the second feature is associated with a second attribute of the second object;
classifying a combination including the first image and the second image into a cluster of image data, wherein the first attribute and the second attribute are identical, and wherein a first real size of the first object is distinct from a second real size of the second object,
generating, based on the cluster of image data, a list of information describing attributes of objects, wherein the list includes information of distinct objects with an identical attribute and in different real sizes;
identifying, using a query image as an input, a third object included in the query image based on a group of images including images including objects with any of a plurality of attributes;
updating, based on the first real size of the first object in the list, a size information of the third object with the first real size of the first object, wherein the third object is classified into the cluster of image data based on a third attribute of the third object being identical to the first attribute of the first object;
calculating, based on the query image including three or more query objects, for each of the query objects of the query image, an actual dimensional ratio based on a size of a query object in the query image and the size information of the third object; and
determining, based on the actual dimensional ratio, a real size of the query object included in the query image.