Skip to main content

Table 3 Hyper-parameters used to train the networks

From: Segmentation of structural parts of rosebush plants with 3D point-based deep learning methods

 

Learning rate

Batch

Decay step/rate

Weight decay

PointNet

0.001

48

30000/0.8

0.005

PointNet++

0.005

12

200000/0.7

None

DGCNN

0.005

12

200000/0.5

None

PointCNN

0.005

8

10000/0.8

1e-8

ShellNet

0.005

12

5000/0.8

1e-8

RIConv

0.005

12

10000/0.8

1e-6