Model card for recon.pretrain.zekun-qi

A ReCon self-supervised pretraining model (masked point transformer with cross-modal contrastive queries). Pretrained on ShapeNet-55.

Model Details

• Model Type: Self-supervised pretraining
• Model Stats:
  • Params (M): 50.8
• Dataset: ShapeNet-55
• Paper: Contrast with Reconstruct: Contrastive 3D Representation Learning Guided by Generative Pretraining
• Converted from: qizekun/ReCon (MIT)
• Library: torch-pointcloud

Install

pip install torch-pointcloud

Usage

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

model, info = tp.create_model(
    "recon.pretrain.zekun-qi",
    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 = collate([sample])

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

Citation

@inproceedings{qi2023recon,
  title     = {Contrast with Reconstruct: Contrastive {3D} Representation Learning Guided by Generative Pretraining},
  author    = {Qi, Zekun and Dong, Runpei and Fan, Guofan and Ge, Zheng and Zhang, Xiangyu and Ma, Kaisheng and Yi, Li},
  booktitle = {ICML},
  year      = {2023}
}

@article{chang2015shapenet,
  author  = {Chang, Angel X. and Funkhouser, Thomas and Guibas, Leonidas and Hanrahan, Pat and Huang, Qixing and Li, Zimo and Savarese, Silvio and Savva, Manolis and Song, Shuran and Su, Hao and Xiao, Jianxiong and Yi, Li and Yu, Fisher},
  title   = {{ShapeNet}: An Information-Rich {3D} Model Repository},
  journal = {arXiv preprint arXiv:1512.03012},
  year    = {2015},
}

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