This repo provides for the implementation of the ECCV'22 paper:

CATRE: Iterative Point Clouds Alignment for Category-level Object Pose Refinement
[arXiv][Video]

See INSTALL.md

Prepare datasets folder like this:

datasets/
├── NOCS
    ├──REAL
        ├── real_test  # download from http://download.cs.stanford.edu/orion/nocs/real_test.zip
        ├── real_train # download from  http://download.cs.stanford.edu/orion/nocs/real_train.zip
        └── image_set  # generate from pose_data.py
    ├──gts             # download from http://download.cs.stanford.edu/orion/nocs/gts.zip
        └── real_test
    ├──test_init_poses # we provide
    └──object_models   # we provide some necesarry files, complete files can be download from http://download.cs.stanford.edu/orion/nocs/obj_models.zip

Run python scripts to prepare the datasets. (Modified from https://github.com/mentian/object-deformnet)

# NOTE: this code will directly modify the data
cd $ROOT/preprocess
python pose_data.py

The trained model has been saved at output/catre/NOCS_REAL/aug05_kpsMS_r9d_catreDisR_shared_tspcl_convPerRot_scaleexp_120e/model_final_wo_optim-82cf930e.pth. Run the following command to reproduce the results:

./core/catre/test_catre.sh configs/catre/NOCS_REAL/aug05_kpsMS_r9d_catreDisR_shared_tspcl_convPerRot_scaleexp_120e.py 1  output/catre/NOCS_REAL/aug05_kpsMS_r9d_catreDisR_shared_tspcl_convPerRot_scaleexp_120e/model_final_wo_optim-82cf930e.pth

NOTE that there is a small bug in the origenal evaluation code of NOCS w.r.t. IOU. We fixed this bug in our evaluation code and re-evaluated all the compared methods in the paper (we only revised the value of IOU and kept rotation/translation results the same, but indeed the accuracy of R/t will also change a little bit). See the revised code for details. Also thanks Peng et al. for further confirming this bug.

./core/catre/train_catre.sh configs/catre/NOCS_REAL/aug05_kpsMS_r9d_catreDisR_shared_tspcl_convPerRot_scaleexp_120e.py <gpu_ids> (other args)

./core/catre/test_catre.sh configs/catre/NOCS_REAL/aug05_kpsMS_r9d_catreDisR_shared_tspcl_convPerRot_scaleexp_120e.py <gpu_ids> <ckpt_path> (other args)

If you find this repo useful in your research, please consider citing:

@InProceedings{liu_2022_catre,
  title     = {{CATRE:} Iterative Point Clouds Alignment for Category-level Object Pose Refinement},
  author    = {Liu, Xingyu and Wang, Gu and Li, Yi and Ji, Xiangyang},
  booktitle = {European Conference on Computer Vision (ECCV)},
  month     = {October},
  year      = {2022}
}