Awesome

SecondPose

Code for "SecondPose: SE(3)-Consistent Dual-Stream Feature Fusion for Category-Level Pose Estimation", Preprint. [Arxiv]

Requirements

The code has been tested with

• python 3.9
• pytorch 1.13.0
• CUDA 11.6

Other dependencies:

requirements.txt

Setup env:

conda create -n secondpose python=3.9
pip install torch==1.13.0+cu116 torchvision==0.14.0+cu116 torchaudio==0.13.0 --extra-index-url https://download.pytorch.org/whl/cu116
cd lib/pointnet2/
pip install .
cd ../sphericalmap_utils/
pip install .
cd ../../
pip install -r requirements.txt
pip install open3d

Data Processing

1. Please refer to the work of Self-DPDN.
2. run data_preprocess.py

Network Training

Train SecondPose for rotation estimation:

python train_geodino.py --gpus 0 --mod r

Train the network of pointnet++ for translation and size estimation:

python train.py --gpus 0  --mod ts 

Evaluation

To test the model, please run:

python test_geodino.py --gpus 0 --test_epoch [YOUR EPOCH]

Model Checkpoints

https://drive.google.com/file/d/1IafKC84XstivhUkm-4rF15KsSHaeBbic/view?usp=sharing

Acknowledgements

Our implementation leverages the code from VI-Net,

License

Our code is released under MIT License (see LICENSE file for details).

Contact

yamei.chen@tum.de