wheat3d_pointcloud

Dataset Summary

wheat3d_pointcloud is a preprocessed version of a 3D point cloud dataset originally stored as .ply files. It contains 3D scans of wheat plant parts that have been cleaned, normalized, and uniformly sampled to a fixed number of points. Each example is saved as a .npy array representing a processed point cloud.

Supported Tasks

• 3D object classification
• Shape completion or reconstruction
• Generative modeling (for example, diffusion models or autoencoders)
• Point cloud denoising and augmentation

Languages

No language data is included (this is a purely geometric dataset).

Dataset Structure

Each example is a .npy file containing a NumPy array of shape (N, 3) for N points with XYZ coordinates.
If color or class information is included, the shape is (N, 6).

Data Fields

• points (np.ndarray): 3D coordinates and optionally RGB or label features of the sampled point cloud
• label (int, optional): class or category of the scan (inferred from folder hierarchy)

Data Splits

No predefined train, validation, or test splits are provided. Please define custom splits as needed for your experiments.

Dataset Creation

1. Iterate over all .ply files in the subdirectories
2. Load each file using Open3D
3. Clean the data (remove NaNs and duplicate points)
4. Normalize the data (center and scale to a unit cube or sphere)
5. Downsample or upsample to a fixed number of points
6. Save the resulting array as a .npy file using NumPy

Citation

@misc{wheatplant3d2025,
  title = {wheat3d_pointcloud: Preprocessed 3D Point Cloud Dataset for ML},
  author = {YourName},
  year = {2025},
  note = {Processed using Open3D, NumPy},
  url = {https://huggingface.co/datasets/yourname/wheat3d_pointcloud}
}