Implement Convpoint with dataset

1

from torch_geometric.datasets import ShapeNet

dataset = ShapeNet(root=‘/tmp/ShapeNet’, categories=[‘Airplane’])

dataset[0]

Data(pos=[2518, 3], y=[2518])

Dear all, i am handling data of point cloud, i want to implement ConvPoint :https://github.com/aboulch/ConvPoint/blob/master/networks/network_classif.py

However, there is only benchmark pretrained model. what is the best way to modify those model and implement network on your owndataset

2

You could copy-paste the linked model and modify it as you wish.
Alternatively you could also use this model as the base class and derive your custom model from it, but it depends on your coding preference I guess.