Automatic Volumetric Segmentation of Additive Manufacturing Defects with 3D U-Net

Published

March 23, 2020

Author(s)

Vivian W. Wong, Max Ferguson, Kincho Law, Yung-Tsun Lee, Paul Witherell

Abstract

Segmentation of defects in additive manufacturing is challenging, due to the poor contrast, small sizes and variation in appearance of defects. Automatic segmentation can however provide quality control for additive manufacturing. Over recent years, 3D convolutional neural networks (3D CNNs) have performed well in the volumetric segmentation of medical images. In this work, we leverage techniques from the medical imaging domain and propose training a 3D U-Net model to automatically segment defects in AM samples. The main contribution of this work is that it is the first to perform 3D volumetric segmentation in AM. We train and test with three variants of the 3D U-Net on an AM dataset, achieving a mean IOU (intersection of union) value of 88.4%.

Proceedings Title

AAAI 2020 SPRING SYMPOSIUM SERIES

Conference Dates

March 23-25, 2020

Conference Location

(Palo Alto-VIRTUAL), CA, US

Conference Title

SSS20 (VIRTUAL) AAAI Spring Symposium on AI in Manufacturing, 3/23-24/2020

Pub Type

Conferences

Download Paper

Local Download

Keywords

Addictive manufacturing, defect segmentation, X-ray computed tomography, Convolutional Neural Networks, 3 D U-Net architecture

Citation

Wong, V. , Ferguson, M. , Law, K. , Lee, Y. and Witherell, P. (2020), Automatic Volumetric Segmentation of Additive Manufacturing Defects with 3D U-Net, AAAI 2020 SPRING SYMPOSIUM SERIES, (Palo Alto-VIRTUAL), CA, US, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=929311 (Accessed March 5, 2025)