US 12,142,067 B2
Learning dataset generation system, learning dataset generation server, and computer readable non-temporary recording medium storing learning dataset generation program
Masafumi Tsutsumi, Osaka (JP)
Assigned to KYOCERA DOCUMENT SOLUTIONS INC., Osaka (JP)
Filed by KYOCERA Document Solutions Inc., Osaka (JP)
Filed on Feb. 19, 2021, as Appl. No. 17/179,461.
Claims priority of application No. 2020-026223 (JP), filed on Feb. 19, 2020.
Prior Publication US 2021/0256252 A1, Aug. 19, 2021
Int. Cl. G06N 20/00 (2019.01); G06F 18/211 (2023.01); G06F 18/214 (2023.01); G06F 18/23 (2023.01); G06F 18/241 (2023.01); G06V 10/22 (2022.01); G06V 10/762 (2022.01); G06V 10/98 (2022.01); G06V 30/40 (2022.01); G06V 30/413 (2022.01); G06V 10/94 (2022.01)
CPC G06V 30/40 (2022.01) [G06F 18/211 (2023.01); G06F 18/214 (2023.01); G06F 18/23 (2023.01); G06F 18/241 (2023.01); G06N 20/00 (2019.01); G06V 10/235 (2022.01); G06V 10/762 (2022.01); G06V 10/987 (2022.01); G06V 30/413 (2022.01); G06V 10/95 (2022.01)] 5 Claims
OG exemplary drawing
 
5. A non-transitory computer-readable recording medium that stores a learning dataset generation program for generating a learning dataset of a document classification inference model that serves as an inference model for classifying a document image, which is an image of a document, by a form thereof and assigning a label to a document image, which is an image of a document, wherein
the learning dataset includes a plurality of labeled document images that have been assigned the label on the basis of the form,
a computer, as a result of executing the learning dataset generation program, divides the plurality of labeled document images into a plurality of clusters by performing clustering on the basis of a feature amount of the labeled document images, selects training data for each of the clusters from among the labeled document images belonging to the cluster, and generates, by learning the training data of all of the clusters, a labeled image classification inference model that serves as an inference model for classifying the labeled document images by the form,
the computer, as a result of executing the learning dataset generation program, performs inference of the labeled document image using the labeled image classification inference model to obtain a degree of certainty with respect to the labeled document image, and confirms the form of the labeled document image in accordance with the degree of certainty with respect to the labeled document image being greater than or equal to a specified value, and
the computer, as a result of executing the learning dataset generation program, provides a UI that causes the labeled document images having the same form as each other in accordance with a result of confirming the form based on the degree of certainty, to be assigned the same label.