抄録
CNN (Convolutional Neural Network) is a powerful method for image classifying tasks. CNN's classifying capability is assessed by using large-scale image dataset such as ImageNet in many papers, but few works on CNN with small-scale dataset have been reported. We have been researching application method of Neural Network for classifying tasks in real-world for years [1]. In this work, we applied CNN to a quality inspection of industrial products and assessed its classifying capacity. Our CNN was trained with 2000 images of eggs taken in a factory, classified the images of almost 89,000 eggs into 6 qualities. Our method of combining multi-angle images into 1 image retained the 3-dimensional features of the object, and improved the classification accuracy to 92.3%. It confirmed that CNN is also effective for the quality inspection of industrial products.
本文言語 | English |
---|---|
ホスト出版物のタイトル | Proceedings - International SoC Design Conference 2017, ISOCC 2017 |
出版社 | Institute of Electrical and Electronics Engineers Inc. |
ページ | 67-68 |
ページ数 | 2 |
ISBN(電子版) | 9781538622858 |
DOI | |
出版ステータス | Published - 2018 5 29 |
イベント | 14th International SoC Design Conference, ISOCC 2017 - Seoul, Korea, Republic of 継続期間: 2017 11 5 → 2017 11 8 |
Other
Other | 14th International SoC Design Conference, ISOCC 2017 |
---|---|
Country | Korea, Republic of |
City | Seoul |
Period | 17/11/5 → 17/11/8 |
ASJC Scopus subject areas
- Hardware and Architecture
- Electrical and Electronic Engineering
- Electronic, Optical and Magnetic Materials