Title:
ABNORMALITY DETECTION METHOD, ABNORMALITY DETECTION DEVICE, AND PROGRAM
Document Type and Number:
WIPO Patent Application WO/2022/250054
Kind Code:
A1
Abstract:
An abnormality detection method for a computer to detect abnormality includes: acquiring (S12) first feature amount data to be output via N convolutional layers (N is an integer of 1 or more) of a convolutional neural network when an image is input (S11) to the convolutional neural network constituted as an encoder; acquiring (S13) second feature amount data to be output via M convolutional layers (M is an integer of 1 or more and M ≠ N is satisfied) of the convolutional neural network, the second feature amount data having a size different from that of the first feature amount data; and detecting (S14) abnormality to the image using feature amounts respectively indicated in the first feature amount data and the second feature amount data having mutually different sizes.
More Like This:
Inventors:
GUDOVSKIY DENIS
ISHIZAKA SHUN (JP)
KOZUKA KAZUKI
ISHIZAKA SHUN (JP)
KOZUKA KAZUKI
Application Number:
PCT/JP2022/021263
Publication Date:
December 01, 2022
Filing Date:
May 24, 2022
Export Citation:
Assignee:
PANASONIC IP CORP AMERICA (US)
International Classes:
G06T7/00
Foreign References:
JP2021081953A | 2021-05-27 |
Attorney, Agent or Firm:
NII, Hiromori et al. (JP)
Download PDF: