US 11,948,091 B2
Image identification apparatus, image identification method, training apparatus, and neural network having sub-neural networks respectively inputted with mutually different data
Takahisa Yamamoto, Kawasaki (JP); and Hiroshi Sato, Kawasaki (JP)
Assigned to CANON KABUSHIKI KAISHA, Tokyo (JP)
Filed by CANON KABUSHIKI KAISHA, Tokyo (JP)
Filed on Jan. 25, 2022, as Appl. No. 17/583,706.
Application 17/583,706 is a continuation of application No. 16/535,289, filed on Aug. 8, 2019, granted, now 11,256,953.
Claims priority of application No. 2018-154214 (JP), filed on Aug. 20, 2018.
Prior Publication US 2022/0148300 A1, May 12, 2022
Int. Cl. G06N 3/084 (2023.01); G06F 18/2111 (2023.01); G06F 18/214 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2023.01); G06V 10/44 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06V 10/96 (2022.01); G06V 40/16 (2022.01)
CPC G06N 3/084 (2013.01) [G06F 18/2111 (2023.01); G06F 18/214 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01); G06V 10/454 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06V 10/96 (2022.01); G06V 40/168 (2022.01); G06V 40/172 (2022.01)] 21 Claims
OG exemplary drawing
 
1. An image identification apparatus, comprising at least one circuit and/or processor, wherein the at least one circuit and/or processor is configured to function as:
an extraction unit configured to extract a feature value of an image from image data using a Neural Network (NN); and
a processing unit configured to identify the image based on the feature value extracted by the extraction unit,
wherein the NN comprises a plurality of calculation layers connected hierarchically,
wherein the NN comprises calculation layers for extracting a feature commonly used for identifying the image in respective regions of the image data,
wherein the NN includes a plurality of sub-neural networks for performing processing of calculation layers after a specific calculation layer among the calculation layers, and wherein mutually different data, each corresponding to a different sub-region of the image data, from an output of the specific calculation layer are respectively inputted to the plurality of sub-neural networks, and
wherein features of facial organs are extracted using the plurality of sub-neural networks.