CPPF: Towards Robust Category-Level 9D Pose Estimation in the Wild — Pretrained Models

Pretrained model checkpoints for CPPF (CVPR 2022), a sim-to-real method for category-level 9D pose estimation trained solely on synthetic ShapeNet models.

• Code: https://github.com/qq456cvb/CPPF
• Project page: https://qq456cvb.github.io/projects/cppf
• Companion dataset (training/eval data): https://huggingface.co/datasets/qq456cvb/CPPF

Contents

One folder per ShapeNet category, each containing:

File	Description
point_encoder_epochbest.pth	Point encoder weights (best epoch)
ppf_encoder_epochbest.pth	PPF encoder weights (best epoch)
.hydra/*.yaml	Hydra config snapshot used for training

Categories: bathtub, bed, bookshelf, bottle, bowl, bowl_reg (regression variant), camera, can, chair, laptop, laptop_aux (auxiliary lid/base segmenter), mug, sofa, table.

Usage

Download and place the category folders under checkpoints/ in the CPPF repository:

pip install -U "huggingface_hub[cli]"
hf download qq456cvb/CPPF --local-dir checkpoints

Citation

@inproceedings{you2022cppf,
  title={CPPF: Towards Robust Category-Level 9D Pose Estimation in the Wild},
  author={You, Yang and Shi, Ruoxi and Wang, Weiming and Lu, Cewu},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year={2022}
}