Defect Classification in Additive Manufacturing Using CNN-Based Vision Processing

TL;DR

This paper explores CNN-based image classification for detecting defects in additive manufacturing and introduces active learning to reduce training data needs, enhancing quality control.

Contribution

It presents a novel CNN-based defect classification method combined with active learning for efficient training data generation in AM.

Findings

CNN achieves high accuracy in defect classification

Active learning reduces data labeling effort

Human-in-the-loop improves model efficiency

Abstract

The development of computer vision and in-situ monitoring using visual sensors allows the collection of large datasets from the additive manufacturing (AM) process. Such datasets could be used with machine learning techniques to improve the quality of AM. This paper examines two scenarios: first, using convolutional neural networks (CNNs) to accurately classify defects in an image dataset from AM and second, applying active learning techniques to the developed classification model. This allows the construction of a human-in-the-loop mechanism to reduce the size of the data required to train and generate training data.