The Application of Convolutional Neural Networks (CNNs) to Recognize Defects in 3D-Printed Parts.

Shengmin Guo,Chang Huang,Hao Wen

doi:10.3390/ma14102575

Shengmin Guo, Chang Huang + Show 1 more

Open Access

https://doi.org/10.3390/ma14102575

Copy DOI

Journal: Materials (Basel, Switzerland)	Publication Date: May 15, 2021
Citations: 14	License type: CC BY 4.0

Affiliation: Louisiana State University

Abstract

Cracks and pores are two common defects in metallic additive manufacturing (AM) parts. In this paper, deep learning-based image analysis is performed for defect (cracks and pores) classification/detection based on SEM images of metallic AM parts. Three different levels of complexities, namely, defect classification, defect detection and defect image segmentation, are successfully achieved using a simple CNN model, the YOLOv4 model and the Detectron2 object detection library, respectively. The tuned CNN model can classify any single defect as either a crack or pore at almost 100% accuracy. The other two models can identify more than 90% of the cracks and pores in the testing images. In addition to the application of static image analysis, defect detection is also successfully applied on a video which mimics the AM process control images. The trained Detectron2 model can identify almost all the pores and cracks that exist in the original video. This study lays a foundation for future in situ process monitoring of the 3D printing process.

Highlights

With the tremendous increase in the computation power of graphics processing units (GPUs), large convolutional neural networks (CNNs) can be trained in a reasonable amount of time
The purpose of this paper is to demonstrate the capability of different CNN models for detecting and classifying cracks and pores, the two common defects in laser beam powder bed fusion (LPBF) parts
0 only after ten epochs of training. Such perfect outcome usually indicates that the model is overfitting, which means the model achieved high accuracy by just remembering each image category

Summary

Introduction

Deep learning-based image analysis is performed for defect (cracks and pores) classification/detection based on SEM images of metallic AM parts. A CNN is a feedforward artificial neural network (ANN) which can accept images directly as an input of the network to avoid complex preprocessing procedures that are carried out in traditional image recognition algorithms [5]. The character of this model requires a large amount of parallel computation, which takes a long time when using central processing units (CPUs) alone.

Objectives

Methods

Results

Conclusion