A MinkUNet point cloud semantic segmentation model (sparse residual U-Net). Trained on SemanticKITTI.
pip install torch-pointcloud
This checkpoint also needs spconv, which needs a build matching your torch and CUDA: see the installation guide.
import torch
import torch_pointcloud as tp
from torch_pointcloud.utils.data import collate
model, info = tp.create_model(
"minkunet18-w16.semantickitti.mit-han-lab",
pretrained=True,
return_info=True,
)
model = model.cuda().eval() # GPU-only kernels
# synthetic sample with the keys a dataset provides
num_points = 8192
sample = {
"pos": torch.randn(num_points, 3),
"intensity": torch.rand(num_points, 1),
"segment": torch.zeros(num_points, dtype=torch.long),
"instance": torch.zeros(num_points, dtype=torch.long),
}
data = info["transform"](sample)
data = collate([data])
data = {key: value.cuda() for key, value in data.items()}
with torch.no_grad():
logits = model(data.get("x"), data["pos_grid"], data["batch"])
with torch.no_grad():
features = model.forward_features(data.get("x"), data["pos_grid"], data["batch"])
model.reset_classifier(num_classes=0)
with torch.no_grad():
features = model(data.get("x"), data["pos_grid"], data["batch"]) # (N, 48)
@inproceedings{choy2019minkowski,
title = {4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks},
author = {Christopher Choy and JunYoung Gwak and Silvio Savarese},
booktitle = {CVPR},
year = {2019}
}
@inproceedings{behley2019semantickitti,
title = {SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences},
author = {Jens Behley and Martin Garbade and Andres Milioto and Jan Quenzel and Sven Behnke and Cyrill Stachniss and Juergen Gall},
booktitle = {ICCV},
year = {2019}
}
@software{dujardin2026pytorchpointcloud,
author = {Arthur Dujardin},
title = {PyTorch PointCloud},
year = {2026},
doi = {10.5281/zenodo.22159632},
url = {https://github.com/arthurdjn/pytorch-pointcloud},
}