dgcnn.scannet20.an-tao

Model card for dgcnn.scannet20.an-tao

A DGCNN point cloud semantic segmentation model (dynamic graph convolution over EdgeConv features). Trained on ScanNet (20 classes).

Model Details

• Model Type: Point cloud semantic segmentation
• Model Stats:
  • Params (M): 1.0
  • Input channels: 6
  • Classes: 20
  • Features: 1216
• Dataset: ScanNet (20 classes)
• Metrics: mIoU 52.89, OA 81.76
• Paper: Dynamic Graph CNN for Learning on Point Clouds
• Converted from: antao97/dgcnn.pytorch (MIT)
• Library: torch-pointcloud

Install

pip install torch-pointcloud

This checkpoint also needs pyg-lib, which needs a build matching your torch and CUDA: see the installation guide.

Usage

import torch
import torch_pointcloud as tp
from torch_pointcloud.utils.data import collate

model, info = tp.create_model(
    "dgcnn.scannet20.an-tao",
    pretrained=True,
    return_info=True,
)
model = model.eval()

# synthetic sample with the keys a dataset provides
num_points = 8192
sample = {
    "pos": torch.randn(num_points, 3),
    "color": torch.rand(num_points, 3) * 255,
    "segment": torch.zeros(num_points, dtype=torch.long),
    "block_center": torch.zeros(3),
    "scene_max": torch.ones(3),
}
data = info["transform"](sample)
data = collate([data])

with torch.no_grad():
    logits = model(data.get("x"), data["pos"], data["batch"])

Feature extraction

with torch.no_grad():
    features = model.forward_features(data.get("x"), data["pos"], data["batch"])

model.reset_classifier(num_classes=0)
with torch.no_grad():
    features = model(data.get("x"), data["pos"], data["batch"])  # (N, 1216)

Citation

@article{wang2019dgcnn,
  title   = {Dynamic Graph CNN for Learning on Point Clouds},
  author  = {Yue Wang and Yongbin Sun and Ziwei Liu and Sanjay E. Sarma and Michael M. Bronstein and Justin M. Solomon},
  journal = {ACM Transactions on Graphics},
  volume  = {38},
  number  = {5},
  year    = {2019}
}

@inproceedings{dai2017scannet,
  title   = {ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes},
  author  = {Angela Dai and Angel X. Chang and Manolis Savva and Maciej Halber and Thomas Funkhouser and Matthias Nießner},
  booktitle = {CVPR},
  year    = {2017}
}

@software{dujardin2026pytorchpointcloud,
  author  = {Arthur Dujardin},
  title   = {PyTorch PointCloud},
  year    = {2026},
  doi     = {10.5281/zenodo.22159632},
  url     = {https://github.com/arthurdjn/pytorch-pointcloud},
}