Table 6 Comparison with relevant research studies.

From: Enhanced detection of surface deformations in LPBF using deep convolutional neural networks and transfer learning from a porosity model

 

Our model

Paper by18

Paper by19

Paper by20

Pre-trained model

VGG-16

VGG-16 ResNet-18

Xception

Data type

Powder bed images

Powder bed images

Acoustic emissions signals

Powder bed images

Image resolution

200 micro-meter/pixel

3.69 micro-meter/pixel

6 micro-meter/pixel

Data size

10593

8904

8000

1356

Target defects

Porosity and surface deformation

Porosity

Balling, lack of fusion, keyholes, conduction mode 

Balling, incomplete spreading, spatter, groove, ridge, scatter powder