Invention Grant
- Patent Title: Inspection system using machine learning to label image segments of defects
-
Application No.: US17258056Application Date: 2019-07-11
-
Publication No.: US11983863B2Publication Date: 2024-05-14
- Inventor: Kazumi Banno , Yuichi Yamagishi , Hiroyuki Kibe , Kunimitsu Toyoshima
- Applicant: Kabushiki Kaisha N-Tech , Kabushiki Kaisha Yakult Honsha , Tohoshoji Kabushiki Kaisha
- Applicant Address: JP Gifu-ken
- Assignee: Kabushiki Kaisha N-Tech
- Current Assignee: Kabushiki Kaisha N-Tech
- Current Assignee Address: JP Gifu-Ken
- Agency: Seager, Tufte & Wickhem LLP
- Priority: JP 18137179 2018.07.20
- International Application: PCT/JP2019/027529 2019.07.11
- International Announcement: WO2020/017427A 2020.01.23
- Date entered country: 2021-01-05
- Main IPC: G06T7/00
- IPC: G06T7/00 ; G06F18/21 ; G06F18/214 ; G06F18/2431 ; G06N5/04 ; G06N20/00 ; G06T7/11

Abstract:
An alpha channel is added in addition to multiple channels forming an image of an item. A second pixel value differing from a first pixel value, which is assigned to a defect-free region, is assigned to a region corresponding to a defect region in the alpha channel. The image including the alpha channel is divided into multiple image segments. A defect label is assigned to the image segment in which the second pixel value is included in the alpha channel, and a non-defect label is assigned to the image segment, in which the second pixel value is not included in the alpha channel. The alpha channel is removed from the image segments to generate labeled image segment data.
Public/Granted literature
Information query