Paper 12623-20
Paper 12623-20
Development of CNN-based asymmetric label smoothing method for visual inspection
On demand | Presented live 28 June 2023 
Abstract
To simultaneously reduce miss and over-detection in visual inspection for defect detection, asymmetric detection techniques practiced by skilled workers were helpful. A CNN-based asymmetric label smoothing method was developed to implement the techniques in a visual inspection.
Presenter
The Graduate School for the Creation of New Photonics Industries (Japan)
Ryohei Hanayama graduated from the University of Tokyo in 2005. After that, he started working for the graduate school for the creation of new photonics industries as an assistant professor. He was promoted to lecturer in 2013. And he currently works as an associate professor for the same school. His research topics are phase shift interferometer, shearing interferometer, and industrial measurement. He got an encouragement award from the Optical Society of Japan in 2005.