TY - JOUR
T1 - Defect detection method using deep convolutional neural network, support vector machine and template matching techniques
AU - Nagata, Fusaomi
AU - Tokuno, Kenta
AU - Mitarai, Kazuki
AU - Otsuka, Akimasa
AU - Ikeda, Takeshi
AU - Ochi, Hiroaki
AU - Watanabe, Keigo
AU - Habib, Maki K.
N1 - Funding Information:
This work was supported by JSPS KAKENHI Grant number 16K06203 and MITSUBISHIPENCIL CO., LTD.
Publisher Copyright:
© 2019, International Society of Artificial Life and Robotics (ISAROB).
PY - 2019/12/1
Y1 - 2019/12/1
N2 - In this paper, a defect detection method using deep convolutional neural network (DCNN), support vector machine (SVM) and template matching techniques is introduced. First, a DCNN for visual inspection is designed and trained using a large number of images to inspect undesirable defects such as crack, burr, protrusion, chipping, spot and fracture phenomena which appear in the manufacturing process of resin molded articles. Then the trained DCNN named sssNet and well-known AlexNet are, respectively, incorporated with two SVMs to classify sample images with high recognition rate into accept as OK category or reject as NG one, in which compressed feature vectors obtained from the DCNNs are used as inputs for the SVMs. The performances of the two types of SVMs with the DCNNs are compared and evaluated through training and classification experiments. Finally, a template matching technique is further proposed to efficiently extract important target areas from original training and test images. This will be able to enhance the reliability and accuracy for defect detection.
AB - In this paper, a defect detection method using deep convolutional neural network (DCNN), support vector machine (SVM) and template matching techniques is introduced. First, a DCNN for visual inspection is designed and trained using a large number of images to inspect undesirable defects such as crack, burr, protrusion, chipping, spot and fracture phenomena which appear in the manufacturing process of resin molded articles. Then the trained DCNN named sssNet and well-known AlexNet are, respectively, incorporated with two SVMs to classify sample images with high recognition rate into accept as OK category or reject as NG one, in which compressed feature vectors obtained from the DCNNs are used as inputs for the SVMs. The performances of the two types of SVMs with the DCNNs are compared and evaluated through training and classification experiments. Finally, a template matching technique is further proposed to efficiently extract important target areas from original training and test images. This will be able to enhance the reliability and accuracy for defect detection.
KW - Deep convolutional neural network (DCNN)
KW - Defect detection system
KW - Support vector machine (SVM)
KW - Template matching
UR - http://www.scopus.com/inward/record.url?scp=85068797129&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85068797129&partnerID=8YFLogxK
U2 - 10.1007/s10015-019-00545-x
DO - 10.1007/s10015-019-00545-x
M3 - Article
AN - SCOPUS:85068797129
VL - 24
SP - 512
EP - 519
JO - Artificial Life and Robotics
JF - Artificial Life and Robotics
SN - 1433-5298
IS - 4
ER -