Anomaly detection for visual quality control of 3D-printed products

Loek Tonnaer, Jiapeng Li, Vladimir Osin, Mike Holenderski, Vlado Menkovski

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

2 Downloads (Pure)

Uittreksel

We present a method for detection of surface defects in images of 3D-printed products that enables automated visual quality control. The data characterising this problem is typically high-dimensional (high-resolution images), imbalanced (defects are relatively rare), and has few labelled examples. We approach these challenges by formulating the problem as probabilistic anomaly detection, where we use Variational Autoencoders (VAE) to estimate the probability density of non-faulty products. We train the VAE in an unsupervised manner on images of non-faulty products only. A successful model will then assign high likelihood to unseen images of non-faulty products, and lower likelihood to images displaying defects.We test this method on anomaly detection scenarios using the MNIST dataset, as well as on images of 3D-printed products. The demonstrated performance is related to the capability of the model to closely estimate the density distribution of the non-faulty (expected) data. For both datasets we present empirical results that the likelihood estimated with a convolutional VAE can separate the normal and anomalous data. Moreover we show how the reconstruction capabilities of VAEs are highly informative for human observers towards localising potential anomalies, which can aid the quality control process.

Originele taal-2Engels
Titel2019 International Joint Conference on Neural Networks, IJCNN 2019
Plaats van productiePiscataway
UitgeverijInstitute of Electrical and Electronics Engineers
Aantal pagina's8
ISBN van elektronische versie978-1-7281-1985-4
DOI's
StatusGepubliceerd - 1 jul 2019
Evenement2019 International Joint Conference on Neural Networks, IJCNN 2019 - Budapest, Hongarije
Duur: 14 jul 201919 jul 2019

Congres

Congres2019 International Joint Conference on Neural Networks, IJCNN 2019
LandHongarije
StadBudapest
Periode14/07/1919/07/19

Vingerafdruk

Quality control
Defects
Surface defects
Image resolution

Citeer dit

Tonnaer, L., Li, J., Osin, V., Holenderski, M., & Menkovski, V. (2019). Anomaly detection for visual quality control of 3D-printed products. In 2019 International Joint Conference on Neural Networks, IJCNN 2019 [8852372] Piscataway: Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/IJCNN.2019.8852372
Tonnaer, Loek ; Li, Jiapeng ; Osin, Vladimir ; Holenderski, Mike ; Menkovski, Vlado. / Anomaly detection for visual quality control of 3D-printed products. 2019 International Joint Conference on Neural Networks, IJCNN 2019. Piscataway : Institute of Electrical and Electronics Engineers, 2019.
@inproceedings{da16eb962d874ffcb1e3d56e355c0c8b,
title = "Anomaly detection for visual quality control of 3D-printed products",
abstract = "We present a method for detection of surface defects in images of 3D-printed products that enables automated visual quality control. The data characterising this problem is typically high-dimensional (high-resolution images), imbalanced (defects are relatively rare), and has few labelled examples. We approach these challenges by formulating the problem as probabilistic anomaly detection, where we use Variational Autoencoders (VAE) to estimate the probability density of non-faulty products. We train the VAE in an unsupervised manner on images of non-faulty products only. A successful model will then assign high likelihood to unseen images of non-faulty products, and lower likelihood to images displaying defects.We test this method on anomaly detection scenarios using the MNIST dataset, as well as on images of 3D-printed products. The demonstrated performance is related to the capability of the model to closely estimate the density distribution of the non-faulty (expected) data. For both datasets we present empirical results that the likelihood estimated with a convolutional VAE can separate the normal and anomalous data. Moreover we show how the reconstruction capabilities of VAEs are highly informative for human observers towards localising potential anomalies, which can aid the quality control process.",
keywords = "anomaly detection, deep learning, variational autoencoder, visual quality control",
author = "Loek Tonnaer and Jiapeng Li and Vladimir Osin and Mike Holenderski and Vlado Menkovski",
year = "2019",
month = "7",
day = "1",
doi = "10.1109/IJCNN.2019.8852372",
language = "English",
booktitle = "2019 International Joint Conference on Neural Networks, IJCNN 2019",
publisher = "Institute of Electrical and Electronics Engineers",
address = "United States",

}

Tonnaer, L, Li, J, Osin, V, Holenderski, M & Menkovski, V 2019, Anomaly detection for visual quality control of 3D-printed products. in 2019 International Joint Conference on Neural Networks, IJCNN 2019., 8852372, Institute of Electrical and Electronics Engineers, Piscataway, Budapest, Hongarije, 14/07/19. https://doi.org/10.1109/IJCNN.2019.8852372

Anomaly detection for visual quality control of 3D-printed products. / Tonnaer, Loek; Li, Jiapeng; Osin, Vladimir; Holenderski, Mike; Menkovski, Vlado.

2019 International Joint Conference on Neural Networks, IJCNN 2019. Piscataway : Institute of Electrical and Electronics Engineers, 2019. 8852372.

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

TY - GEN

T1 - Anomaly detection for visual quality control of 3D-printed products

AU - Tonnaer, Loek

AU - Li, Jiapeng

AU - Osin, Vladimir

AU - Holenderski, Mike

AU - Menkovski, Vlado

PY - 2019/7/1

Y1 - 2019/7/1

N2 - We present a method for detection of surface defects in images of 3D-printed products that enables automated visual quality control. The data characterising this problem is typically high-dimensional (high-resolution images), imbalanced (defects are relatively rare), and has few labelled examples. We approach these challenges by formulating the problem as probabilistic anomaly detection, where we use Variational Autoencoders (VAE) to estimate the probability density of non-faulty products. We train the VAE in an unsupervised manner on images of non-faulty products only. A successful model will then assign high likelihood to unseen images of non-faulty products, and lower likelihood to images displaying defects.We test this method on anomaly detection scenarios using the MNIST dataset, as well as on images of 3D-printed products. The demonstrated performance is related to the capability of the model to closely estimate the density distribution of the non-faulty (expected) data. For both datasets we present empirical results that the likelihood estimated with a convolutional VAE can separate the normal and anomalous data. Moreover we show how the reconstruction capabilities of VAEs are highly informative for human observers towards localising potential anomalies, which can aid the quality control process.

AB - We present a method for detection of surface defects in images of 3D-printed products that enables automated visual quality control. The data characterising this problem is typically high-dimensional (high-resolution images), imbalanced (defects are relatively rare), and has few labelled examples. We approach these challenges by formulating the problem as probabilistic anomaly detection, where we use Variational Autoencoders (VAE) to estimate the probability density of non-faulty products. We train the VAE in an unsupervised manner on images of non-faulty products only. A successful model will then assign high likelihood to unseen images of non-faulty products, and lower likelihood to images displaying defects.We test this method on anomaly detection scenarios using the MNIST dataset, as well as on images of 3D-printed products. The demonstrated performance is related to the capability of the model to closely estimate the density distribution of the non-faulty (expected) data. For both datasets we present empirical results that the likelihood estimated with a convolutional VAE can separate the normal and anomalous data. Moreover we show how the reconstruction capabilities of VAEs are highly informative for human observers towards localising potential anomalies, which can aid the quality control process.

KW - anomaly detection

KW - deep learning

KW - variational autoencoder

KW - visual quality control

UR - http://www.scopus.com/inward/record.url?scp=85073213843&partnerID=8YFLogxK

U2 - 10.1109/IJCNN.2019.8852372

DO - 10.1109/IJCNN.2019.8852372

M3 - Conference contribution

AN - SCOPUS:85073213843

BT - 2019 International Joint Conference on Neural Networks, IJCNN 2019

PB - Institute of Electrical and Electronics Engineers

CY - Piscataway

ER -

Tonnaer L, Li J, Osin V, Holenderski M, Menkovski V. Anomaly detection for visual quality control of 3D-printed products. In 2019 International Joint Conference on Neural Networks, IJCNN 2019. Piscataway: Institute of Electrical and Electronics Engineers. 2019. 8852372 https://doi.org/10.1109/IJCNN.2019.8852372