https://github.com/jedrazb/vtk-to-pytorch-geometric

PyTorch Geometric point cloud representation for vtk 3D objects
https://github.com/jedrazb/vtk-to-pytorch-geometric

dataset point-cloud pytorch-geometric vtk vtk-library

Last synced: about 1 month ago
JSON representation

PyTorch Geometric point cloud representation for vtk 3D objects

• Host: GitHub
• URL: https://github.com/jedrazb/vtk-to-pytorch-geometric
• Owner: jedrazb
• Created: 2020-06-29T21:19:42.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-06-29T22:52:59.000Z (over 4 years ago)
• Last Synced: 2024-10-30T00:05:59.590Z (3 months ago)
• Topics: dataset, point-cloud, pytorch-geometric, vtk, vtk-library
• Language: Python
• Homepage:
• Size: 6.84 KB
• Stars: 2
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# vtk-to-pytorch-geometric

Generate point cloud representation of the `.vtk` 3d object. This representation is compatible with PyTorch Geometric library.

## How to

### Read your data

First implement `read_data_objects` function in `utils.py`. You need to represent each 3d object as a dict with following attributes:

```python

{

  'path': (string): absolute path to this .vtk object

  'y': (int): label of the object

  ... and optional attributes

}

```

Note that you can pass more attributes to the dict. Those will be automatically passed to the PyTorch Geometric abstraction as well. 

### Generate and save your PyTorch Geometric Dataset

To generate the dataset use `generate.py` script

```bash

python generate.py \

--path /path/to/dataset \

--num_points 1024

```

Args:

- `path` arguments means where your dataset will be stored. Dataset generated to `path` just once. All instances of Dataset object (e.g. in your train code) with existing path will just read the data instead of processing it.

- `num_points` Number of points uniformly sampled from the shape surface