US 11,055,584 B2
Image processing apparatus, image processing method, and non-transitory computer-readable storage medium that perform class identification of an input image using a discriminator that has undergone learning to perform class identification at different granularities
Masato Aoba, Tokyo (JP); and Yasuhiro Komori, Tokyo (JP)
Assigned to CANON KABUSHIKI KAISHA, Tokyo (JP)
Filed by CANON KABUSHIKI KAISHA, Tokyo (JP)
Filed on Jul. 6, 2018, as Appl. No. 16/28,633.
Claims priority of application No. JP2017-134949 (JP), filed on Jul. 10, 2017.
Prior Publication US 2019/0012582 A1, Jan. 10, 2019
Int. Cl. G06K 9/62 (2006.01); G06K 9/46 (2006.01); G06K 9/68 (2006.01); G06T 7/136 (2017.01); G06K 9/00 (2006.01)
CPC G06K 9/6857 (2013.01) [G06K 9/00664 (2013.01); G06K 9/00671 (2013.01); G06K 9/4642 (2013.01); G06K 9/6268 (2013.01); G06K 9/6281 (2013.01); G06T 7/136 (2017.01); G06T 2207/20016 (2013.01); G06T 2207/20081 (2013.01)] 17 Claims
OG exemplary drawing
 
1. An image processing apparatus comprising:
a discriminator that has undergone learning based on image feature of a first image to perform class identification at each of a plurality of granularities, wherein:
the plurality of granularities include a first granularity and a second granularity coarser than the first granularity, and
a class of interest, among classes defined with the second granularity, is an integrated class obtained by integrating some classes corresponding to the class of interest, among classes defined with the first granularity; and
one or more processors configured to execute a plurality of tasks, including:
an acquisition task that acquires an input image;
a selection task that selects, based on image information of the input image, a granularity used for class identification of the input image from the plurality of granularities; and
an identification task that, based on an output obtained by inputting an image feature of the input image and the granularity selected by the selection task to the discriminator, performs class identification of the input image at the granularity selected by the selection task.