README.md

---
license: mit
library_name: torch-pointcloud
tags:
- point-cloud
- 3d
- pytorch
- torch-pointcloud
- point-bert
- classification
datasets:
- scanobjectnn
base_model: torch-pointcloud/point-bert-base.pretrain.xumin-yu
model-index:
- name: point-bert-base.scanobjectnn-objonly.xumin-yu
  results:
  - task:
      type: point-cloud-classification
    dataset:
      name: ScanObjectNN (OBJ_ONLY)
      type: scanobjectnn
    metrics:
    - name: OA
      type: accuracy
      value: 88.12
    - name: mAcc
      type: accuracy
      value: 85.93
---

# Model card for point-bert-base.scanobjectnn-objonly.xumin-yu

A Point-BERT point cloud classification model (masked point modeling transformer). Trained on ScanObjectNN (OBJ_ONLY).

## Model Details

- **Model Type:** Point cloud classification
- **Model Stats:**
  - Params (M): 22.1
  - Classes: 40
  - Features: 768
- **Dataset:** ScanObjectNN (OBJ_ONLY)
- **Metrics:** OA 88.12, mAcc 85.93
- **Paper:** [Point-BERT: Pre-training 3D Point Cloud Transformers with Masked Point Modeling](https://arxiv.org/abs/2111.14819)
- **Converted from:** [Julie-tang00/Point-BERT](https://github.com/Julie-tang00/Point-BERT) (MIT)
- **Library:** [torch-pointcloud](https://github.com/arthurdjn/pytorch-pointcloud)

## Install

```bash
pip install torch-pointcloud
```

## Usage

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

model, info = tp.create_model(
    "point-bert-base.scanobjectnn-objonly.xumin-yu",
    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),
}
data = info["transform"](sample)
data = collate([data])

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

## Feature extraction

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

model.reset_classifier(num_classes=0)
with torch.no_grad():
    embeddings = model(data.get("x"), data["pos"], data["batch"])  # (B, 768)
```

## Citation

```bibtex
@inproceedings{yu2022pointbert,
  title   = {Point-BERT: Pre-training 3D Point Cloud Transformers with Masked Point Modeling},
  author  = {Xumin Yu and Lulu Tang and Yongming Rao and Tiejun Huang and Jie Zhou and Jiwen Lu},
  booktitle = {CVPR},
  year    = {2022}
}

@inproceedings{uy2019scanobjectnn,
  title   = {Revisiting Point Cloud Classification: A New Benchmark Dataset and Classification Model on Real-World Data},
  author  = {Mikaela Angelina Uy and Quang-Hieu Pham and Binh-Son Hua and Duc Thanh Nguyen and Sai-Kit Yeung},
  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},
}
```