GanoDIP - GAN Anomaly Detection through Intermediate Patches: a PCBA Manufacturing Case

Arnaud Bougaham; Adrien Bibal; Isabelle Linden; Benoît Frénay

GanoDIP - GAN Anomaly Detection through Intermediate Patches: a PCBA Manufacturing Case

Arnaud Bougaham, Adrien Bibal, Isabelle Linden, Benoît Frénay

Research output: Contribution to conference › Paper › peer-review

Abstract

Industry 4.0 and recent deep learning progress make it possible to solve problems that traditional methods could not. This is the case for anomaly detection that received a particular attention from the machine learning community, and resulted in a use of generative adversarial networks (GANs). In this work, we propose to use intermediate patches for the inference step, after a WGAN training procedure suitable for highly imbalanced datasets, to make the anomaly detection possible on full size Printed Circuit Board Assembly (PCBA) images. We therefore show that our technique can be used to support or replace actual industrial image processing algorithms, as well as to avoid a waste of time for industries.

Original language	English
Pages	1-14
Number of pages	14
Publication status	Published - 29 Sept 2021
Event	Third International Workshop on Learning with Imbalanced Domains: Theory and Applications; co-located with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD) 2021. - Virtual (formerly Bilbao-Spain), Bilbao, Spain Duration: 17 Sept 2021 → 17 Sept 2021 https://lidta.dcc.fc.up.pt/

Workshop

Workshop	Third International Workshop on Learning with Imbalanced Domains: Theory and Applications; co-located with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD) 2021.
Abbreviated title	LIDTA 2021
Country/Territory	Spain
City	Bilbao
Period	17/09/21 → 17/09/21
Internet address	https://lidta.dcc.fc.up.pt/

UN SDGs

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

Access to Document

https://proceedings.mlr.press/v154/bougaham21a/bougaham21a.pdf

Cite this

Bougaham, A., Bibal, A., Linden, I., & Frénay, B. (2021). GanoDIP - GAN Anomaly Detection through Intermediate Patches: a PCBA Manufacturing Case. 1-14. Paper presented at Third International Workshop on Learning with Imbalanced Domains: Theory and Applications; co-located with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD) 2021., Bilbao, Spain. https://proceedings.mlr.press/v154/bougaham21a/bougaham21a.pdf

Bougaham, Arnaud ; Bibal, Adrien ; Linden, Isabelle et al. / GanoDIP - GAN Anomaly Detection through Intermediate Patches : a PCBA Manufacturing Case. Paper presented at Third International Workshop on Learning with Imbalanced Domains: Theory and Applications; co-located with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD) 2021., Bilbao, Spain.14 p.

@conference{b4af075592d947fab1fae0e6e5ea5944,

title = "GanoDIP - GAN Anomaly Detection through Intermediate Patches: a PCBA Manufacturing Case",

abstract = "Industry 4.0 and recent deep learning progress make it possible to solve problems that traditional methods could not. This is the case for anomaly detection that received a particular attention from the machine learning community, and resulted in a use of generative adversarial networks (GANs). In this work, we propose to use intermediate patches for the inference step, after a WGAN training procedure suitable for highly imbalanced datasets, to make the anomaly detection possible on full size Printed Circuit Board Assembly (PCBA) images. We therefore show that our technique can be used to support or replace actual industrial image processing algorithms, as well as to avoid a waste of time for industries. ",

author = "Arnaud Bougaham and Adrien Bibal and Isabelle Linden and Beno{\^i}t Fr{\'e}nay",

year = "2021",

month = sep,

day = "29",

language = "English",

pages = "1--14",

note = "Third International Workshop on Learning with Imbalanced Domains: Theory and Applications; co-located with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD) 2021., LIDTA 2021 ; Conference date: 17-09-2021 Through 17-09-2021",

url = "https://lidta.dcc.fc.up.pt/",

}

Bougaham, A , Bibal, A , Linden, I & Frénay, B 2021, 'GanoDIP - GAN Anomaly Detection through Intermediate Patches: a PCBA Manufacturing Case', Paper presented at Third International Workshop on Learning with Imbalanced Domains: Theory and Applications; co-located with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD) 2021., Bilbao, Spain, 17/09/21 - 17/09/21 pp. 1-14. <https://proceedings.mlr.press/v154/bougaham21a/bougaham21a.pdf>

GanoDIP - GAN Anomaly Detection through Intermediate Patches: a PCBA Manufacturing Case. / Bougaham, Arnaud ; Bibal, Adrien ; Linden, Isabelle et al.
2021. 1-14 Paper presented at Third International Workshop on Learning with Imbalanced Domains: Theory and Applications; co-located with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD) 2021., Bilbao, Spain.

Research output: Contribution to conference › Paper › peer-review

TY - CONF

T1 - GanoDIP - GAN Anomaly Detection through Intermediate Patches

T2 - Third International Workshop on Learning with Imbalanced Domains: Theory and Applications; co-located with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD) 2021.

AU - Bougaham, Arnaud

AU - Bibal, Adrien

AU - Linden, Isabelle

AU - Frénay, Benoît

PY - 2021/9/29

Y1 - 2021/9/29

N2 - Industry 4.0 and recent deep learning progress make it possible to solve problems that traditional methods could not. This is the case for anomaly detection that received a particular attention from the machine learning community, and resulted in a use of generative adversarial networks (GANs). In this work, we propose to use intermediate patches for the inference step, after a WGAN training procedure suitable for highly imbalanced datasets, to make the anomaly detection possible on full size Printed Circuit Board Assembly (PCBA) images. We therefore show that our technique can be used to support or replace actual industrial image processing algorithms, as well as to avoid a waste of time for industries.

AB - Industry 4.0 and recent deep learning progress make it possible to solve problems that traditional methods could not. This is the case for anomaly detection that received a particular attention from the machine learning community, and resulted in a use of generative adversarial networks (GANs). In this work, we propose to use intermediate patches for the inference step, after a WGAN training procedure suitable for highly imbalanced datasets, to make the anomaly detection possible on full size Printed Circuit Board Assembly (PCBA) images. We therefore show that our technique can be used to support or replace actual industrial image processing algorithms, as well as to avoid a waste of time for industries.

M3 - Paper

SP - 1

EP - 14

Y2 - 17 September 2021 through 17 September 2021

ER -

Bougaham A , Bibal A , Linden I , Frénay B. GanoDIP - GAN Anomaly Detection through Intermediate Patches: a PCBA Manufacturing Case. 2021. Paper presented at Third International Workshop on Learning with Imbalanced Domains: Theory and Applications; co-located with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD) 2021., Bilbao, Spain.

GanoDIP - GAN Anomaly Detection through Intermediate Patches: a PCBA Manufacturing Case

Abstract

Workshop

UN SDGs

Access to Document

Fingerprint

Cite this