JSAI2020

Presentation information

Interactive Session

[4Rin1] Interactive 2

Fri. Jun 12, 2020 9:00 AM - 10:40 AM Room R01 (jsai2020online-2-33)

[4Rin1-60] Application of deep learning and preprocessing to the apperance check of low tank of toilet

〇MANABU UMEDA1, NAOKI KOGA2 (1.LIXIL Co., 2.Computermind Co.)

Keywords:Deep learning, Sanitary ware, Annotation, Picture image

The appearance check of a low tank of toilet is carried out with naked eyes of a skilled craftsman. It is diffucult to detect small defects with high accuracy at glazed surfaces with naked eyes. Therefore, the automated appearance check with high accuracy is hoped to be developed. In this study, the deep learing is applied to detect defaults. A great variety of samples are needed for the training to obtain enough generalization. However, it is difficult to get enough samples with its low defect rate. To deal with this problem, an annotation and preprocessing are contrived for the deep learing. This brings the improvement of the accuracy for the practical use at the factory.

Authentication for paper PDF access

A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.

Password