Title:
TEACHER DATA GENERATION METHOD, TRAINED MODEL GENERATION METHOD, DEVICE, RECORDING MEDIUM, PROGRAM, AND INFORMATION PROCESSING DEVICE
Document Type and Number:
WIPO Patent Application WO/2021/090587
Kind Code:
A1
Abstract:
According to the present invention, teacher data is generated on the basis of normal content and abnormal content generated from the normal content. Provided is a teacher data generation method which generates teacher data used for generating a trained model for determining whether an inspection target has an abnormality, said method including receiving normal content pertaining to the inspection target and abnormal content generated from the normal content, and generating teacher data on the basis of a set of the normal content and one or more pieces of the abnormal content.
Inventors:
WANG XIAODONG (JP)
Application Number:
PCT/JP2020/035212
Publication Date:
May 14, 2021
Filing Date:
September 17, 2020
Export Citation:
Assignee:
LEADER ELECTRONICS (JP)
International Classes:
G06N20/00; G06N3/08
Foreign References:
| JPH08320251A | 1996-12-03 | |||
| JP2018124937A | 2018-08-09 | |||
| JP2018206262A | 2018-12-27 | |||
| JP2018120300A | 2018-08-02 |
Other References:
DAISUKELAB: "Defect discovery! Deep Metric Learning application to MVTec anomaly detection datasets", QIITA, 17 September 2019 (2019-09-17), XP055823554, Retrieved from the Internet [retrieved on 20210704]
Attorney, Agent or Firm:
YAMAMOTO, Osamu et al. (JP)
Download PDF:
Previous Patent: PAPER SHEET FEEDING APPARATUS AND BOX MAKING MACHINE
Next Patent: SEAT AND SEAT PRODUCTION METHOD
Next Patent: SEAT AND SEAT PRODUCTION METHOD