Robust Inspection of Micro-LED Chip Defects Using Unsupervised Anomaly Detection

Sooill Park; Jong Hwan Ko

doi:10.1109/ICTC52510.2021.9620801

Robust Inspection of Micro-LED Chip Defects Using Unsupervised Anomaly Detection

Sooill Park
, Jong Hwan Ko

Department of Electronic and Electrical Engineering

Sungkyunkwan University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

15 Scopus citations

Abstract

Inspection of defects in light emitting diode (LED) chips have been studied to reduce manufacturing cost. Recent studies proposed visual defect inspection methods based on supervised learning of deep neural networks. However, they require datasets with the ground-truth label of each chip, which accompanies prohibitively laborious tasks. In addition, they require class balanced datasets, which is difficult to obtain in an actual industrial process. In order to tackle these limitations, this paper proposes an unsupervised learning based inspection method using anomaly detection that requires no labeled data. On the micro-LED dataset, we demonstrate that our method outperforms previous anomaly detection methods. We achieve 95.82% AUROC result, which is 20.87% higher than convolutional autoencoder and 0.67% higher than Deep SVDD.

Original language	English
Title of host publication	ICTC 2021 - 12th International Conference on ICT Convergence
Subtitle of host publication	Beyond the Pandemic Era with ICT Convergence Innovation
Publisher	IEEE Computer Society
Pages	1841-1843
Number of pages	3
ISBN (Electronic)	9781665423830
DOIs	https://doi.org/10.1109/ICTC52510.2021.9620801
State	Published - 2021
Event	12th International Conference on Information and Communication Technology Convergence, ICTC 2021 - Jeju Island, Korea, Republic of Duration: 20 Oct 2021 → 22 Oct 2021

Publication series

Name	International Conference on ICT Convergence
Volume	2021-October
ISSN (Print)	2162-1233
ISSN (Electronic)	2162-1241

Conference

Conference	12th International Conference on Information and Communication Technology Convergence, ICTC 2021
Country/Territory	Korea, Republic of
City	Jeju Island
Period	20/10/21 → 22/10/21

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 9 Industry, Innovation, and Infrastructure

Keywords

Anomaly Detection
Deep Learning
Defect Detection
Micro-LED
Visual Inspection

Access to Document

10.1109/ICTC52510.2021.9620801

Cite this

Park, S., & Ko, J. H. (2021). Robust Inspection of Micro-LED Chip Defects Using Unsupervised Anomaly Detection. In ICTC 2021 - 12th International Conference on ICT Convergence: Beyond the Pandemic Era with ICT Convergence Innovation (pp. 1841-1843). (International Conference on ICT Convergence; Vol. 2021-October). IEEE Computer Society. https://doi.org/10.1109/ICTC52510.2021.9620801

@inproceedings{e0412d5f27d24af2a737f2d33f84c46b,

title = "Robust Inspection of Micro-LED Chip Defects Using Unsupervised Anomaly Detection",

abstract = "Inspection of defects in light emitting diode (LED) chips have been studied to reduce manufacturing cost. Recent studies proposed visual defect inspection methods based on supervised learning of deep neural networks. However, they require datasets with the ground-truth label of each chip, which accompanies prohibitively laborious tasks. In addition, they require class balanced datasets, which is difficult to obtain in an actual industrial process. In order to tackle these limitations, this paper proposes an unsupervised learning based inspection method using anomaly detection that requires no labeled data. On the micro-LED dataset, we demonstrate that our method outperforms previous anomaly detection methods. We achieve 95.82\% AUROC result, which is 20.87\% higher than convolutional autoencoder and 0.67\% higher than Deep SVDD.",

keywords = "Anomaly Detection, Deep Learning, Defect Detection, Micro-LED, Visual Inspection",

author = "Sooill Park and Ko, \{Jong Hwan\}",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 12th International Conference on Information and Communication Technology Convergence, ICTC 2021 ; Conference date: 20-10-2021 Through 22-10-2021",

year = "2021",

doi = "10.1109/ICTC52510.2021.9620801",

language = "English",

series = "International Conference on ICT Convergence",

publisher = "IEEE Computer Society",

pages = "1841--1843",

booktitle = "ICTC 2021 - 12th International Conference on ICT Convergence",

}

Park, S & Ko, JH 2021, Robust Inspection of Micro-LED Chip Defects Using Unsupervised Anomaly Detection. in ICTC 2021 - 12th International Conference on ICT Convergence: Beyond the Pandemic Era with ICT Convergence Innovation. International Conference on ICT Convergence, vol. 2021-October, IEEE Computer Society, pp. 1841-1843, 12th International Conference on Information and Communication Technology Convergence, ICTC 2021, Jeju Island, Korea, Republic of, 20/10/21. https://doi.org/10.1109/ICTC52510.2021.9620801

Robust Inspection of Micro-LED Chip Defects Using Unsupervised Anomaly Detection. / Park, Sooill; Ko, Jong Hwan.
ICTC 2021 - 12th International Conference on ICT Convergence: Beyond the Pandemic Era with ICT Convergence Innovation. IEEE Computer Society, 2021. p. 1841-1843 (International Conference on ICT Convergence; Vol. 2021-October).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Robust Inspection of Micro-LED Chip Defects Using Unsupervised Anomaly Detection

AU - Park, Sooill

AU - Ko, Jong Hwan

PY - 2021

Y1 - 2021

N2 - Inspection of defects in light emitting diode (LED) chips have been studied to reduce manufacturing cost. Recent studies proposed visual defect inspection methods based on supervised learning of deep neural networks. However, they require datasets with the ground-truth label of each chip, which accompanies prohibitively laborious tasks. In addition, they require class balanced datasets, which is difficult to obtain in an actual industrial process. In order to tackle these limitations, this paper proposes an unsupervised learning based inspection method using anomaly detection that requires no labeled data. On the micro-LED dataset, we demonstrate that our method outperforms previous anomaly detection methods. We achieve 95.82% AUROC result, which is 20.87% higher than convolutional autoencoder and 0.67% higher than Deep SVDD.

AB - Inspection of defects in light emitting diode (LED) chips have been studied to reduce manufacturing cost. Recent studies proposed visual defect inspection methods based on supervised learning of deep neural networks. However, they require datasets with the ground-truth label of each chip, which accompanies prohibitively laborious tasks. In addition, they require class balanced datasets, which is difficult to obtain in an actual industrial process. In order to tackle these limitations, this paper proposes an unsupervised learning based inspection method using anomaly detection that requires no labeled data. On the micro-LED dataset, we demonstrate that our method outperforms previous anomaly detection methods. We achieve 95.82% AUROC result, which is 20.87% higher than convolutional autoencoder and 0.67% higher than Deep SVDD.

KW - Anomaly Detection

KW - Deep Learning

KW - Defect Detection

KW - Micro-LED

KW - Visual Inspection

UR - https://www.scopus.com/pages/publications/85122954666

U2 - 10.1109/ICTC52510.2021.9620801

DO - 10.1109/ICTC52510.2021.9620801

M3 - Conference contribution

AN - SCOPUS:85122954666

T3 - International Conference on ICT Convergence

SP - 1841

EP - 1843

BT - ICTC 2021 - 12th International Conference on ICT Convergence

PB - IEEE Computer Society

T2 - 12th International Conference on Information and Communication Technology Convergence, ICTC 2021

Y2 - 20 October 2021 through 22 October 2021

ER -

Robust Inspection of Micro-LED Chip Defects Using Unsupervised Anomaly Detection

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this