3D Deep Learning Models

Dynamic-Graph CNN

Flux3D.DGCNNType
DGCNN(num_classes::Int=10, K::Int=10, npoints::Int=1024)

Flux implementation of Dynamic-Graph CNN classification model.

Fields:

  • num_classes - Number of classes in dataset.
  • K - k nearest neighbour to be used EdgeConv.
  • npoints - Number of points in input PointCloud.
source

PointNet

Flux3D.PointNetType
PointNet(num_classes::Int=10, hidden_dims::Int=64)

Flux implementation of PointNet classification model.

Fields:

  • num_classes - Number of classes in dataset.
  • hidden_dims - Hiddem dimension in PointNet model.
source