Defect Detection in Printed Circuit Boards Using Semi-Supervised Learning
Defect inspection is essential in the semiconductor industry to fabricate printed circuit boards (PCBs) with minimum defect rates. However, conventional inspection systems are labor-intensive and time-consuming. In this study, a semi-supervised learning (SSL)-based model called PCB_SS was developed....
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2023-03-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/23/6/3246 |
_version_ | 1797609097621667840 |
---|---|
author | Thi Tram Anh Pham Do Kieu Trang Thoi Hyohoon Choi Suhyun Park |
author_facet | Thi Tram Anh Pham Do Kieu Trang Thoi Hyohoon Choi Suhyun Park |
author_sort | Thi Tram Anh Pham |
collection | DOAJ |
description | Defect inspection is essential in the semiconductor industry to fabricate printed circuit boards (PCBs) with minimum defect rates. However, conventional inspection systems are labor-intensive and time-consuming. In this study, a semi-supervised learning (SSL)-based model called PCB_SS was developed. It was trained using labeled and unlabeled images under two different augmentations. Training and test PCB images were acquired using automatic final vision inspection systems. The PCB_SS model outperformed a completely supervised model trained using only labeled images (PCB_FS). The performance of the PCB_SS model was more robust than that of the PCB_FS model when the number of labeled data is limited or comprises incorrectly labeled data. In an error-resilience test, the proposed PCB_SS model maintained stable accuracy (error increment of less than 0.5%, compared with 4% for PCB_FS) for noisy training data (with as much as 9.0% of the data labeled incorrectly). The proposed model also showed superior performance when comparing machine-learning and deep-learning classifiers. The unlabeled data utilized in the PCB_SS model helped with the generalization of the deep-learning model and improved its performance for PCB defect detection. Thus, the proposed method alleviates the burden of the manual labeling process and provides a rapid and accurate automatic classifier for PCB inspections. |
first_indexed | 2024-03-11T05:55:47Z |
format | Article |
id | doaj.art-e489d894006541af82553e9ca97cb6da |
institution | Directory Open Access Journal |
issn | 1424-8220 |
language | English |
last_indexed | 2024-03-11T05:55:47Z |
publishDate | 2023-03-01 |
publisher | MDPI AG |
record_format | Article |
series | Sensors |
spelling | doaj.art-e489d894006541af82553e9ca97cb6da2023-11-17T13:48:01ZengMDPI AGSensors1424-82202023-03-01236324610.3390/s23063246Defect Detection in Printed Circuit Boards Using Semi-Supervised LearningThi Tram Anh Pham0Do Kieu Trang Thoi1Hyohoon Choi2Suhyun Park3Department of Electronic and Electrical Engineering, Ewha Womans University, Seoul 03760, Republic of KoreaDepartment of Electronic and Electrical Engineering, Ewha Womans University, Seoul 03760, Republic of KoreaPixel Inc., Pyeongtaek 17708, Republic of KoreaDepartment of Electronic and Electrical Engineering, Ewha Womans University, Seoul 03760, Republic of KoreaDefect inspection is essential in the semiconductor industry to fabricate printed circuit boards (PCBs) with minimum defect rates. However, conventional inspection systems are labor-intensive and time-consuming. In this study, a semi-supervised learning (SSL)-based model called PCB_SS was developed. It was trained using labeled and unlabeled images under two different augmentations. Training and test PCB images were acquired using automatic final vision inspection systems. The PCB_SS model outperformed a completely supervised model trained using only labeled images (PCB_FS). The performance of the PCB_SS model was more robust than that of the PCB_FS model when the number of labeled data is limited or comprises incorrectly labeled data. In an error-resilience test, the proposed PCB_SS model maintained stable accuracy (error increment of less than 0.5%, compared with 4% for PCB_FS) for noisy training data (with as much as 9.0% of the data labeled incorrectly). The proposed model also showed superior performance when comparing machine-learning and deep-learning classifiers. The unlabeled data utilized in the PCB_SS model helped with the generalization of the deep-learning model and improved its performance for PCB defect detection. Thus, the proposed method alleviates the burden of the manual labeling process and provides a rapid and accurate automatic classifier for PCB inspections.https://www.mdpi.com/1424-8220/23/6/3246defect inspectionnoisy trainingprinted circuit boardsemi-supervised learning |
spellingShingle | Thi Tram Anh Pham Do Kieu Trang Thoi Hyohoon Choi Suhyun Park Defect Detection in Printed Circuit Boards Using Semi-Supervised Learning Sensors defect inspection noisy training printed circuit board semi-supervised learning |
title | Defect Detection in Printed Circuit Boards Using Semi-Supervised Learning |
title_full | Defect Detection in Printed Circuit Boards Using Semi-Supervised Learning |
title_fullStr | Defect Detection in Printed Circuit Boards Using Semi-Supervised Learning |
title_full_unstemmed | Defect Detection in Printed Circuit Boards Using Semi-Supervised Learning |
title_short | Defect Detection in Printed Circuit Boards Using Semi-Supervised Learning |
title_sort | defect detection in printed circuit boards using semi supervised learning |
topic | defect inspection noisy training printed circuit board semi-supervised learning |
url | https://www.mdpi.com/1424-8220/23/6/3246 |
work_keys_str_mv | AT thitramanhpham defectdetectioninprintedcircuitboardsusingsemisupervisedlearning AT dokieutrangthoi defectdetectioninprintedcircuitboardsusingsemisupervisedlearning AT hyohoonchoi defectdetectioninprintedcircuitboardsusingsemisupervisedlearning AT suhyunpark defectdetectioninprintedcircuitboardsusingsemisupervisedlearning |