jibweb / cope

About the repository branch

This is a cleaner version of COPE, no shared data loader and evaluation; camera instrinsics are passed as input to the network, allowing different intrinsics during training and testing. Repository for generating the paper results is available at https://github.com/sThalham/COPE/tree/cope_WACV

COPE: End-to-end trainable Constant Runtime Object Pose Estimation

Stefan Thalhammer, Timothy Patten, Markus Vincze,

Accepted for publication at WACV: Winter Conference on Applications in Computer Vision, 2023, algorithms track

[Paper]

Citation

Please cite the paper if you are using the code:

@inproceedings{thalhammer2023cope,
title= {COPE: End-to-end trainable constant runtime object pose estimation}
author={S. {Thalhammer} and T. {Patten} and M. {Vincze}},
journal={arXiv preprint arXiv:2208.08807},
year={2022}}

Installation

git clone https://github.com/sThalham/COPE.git
python3 -m pip install opencv-python==4.4.0.40
python3 -m pip install pillow
python3 -m pip install matplotlib
python3 -m pip install transforms3d
python3 -m pip install glumpy
python3 -m pip install open3d-python
python3 -m pip install PyOpenGL
python3 -m pip install imgaug

Alternatively, use the provided Dockerfile to deploy a Docker container that satisfies the version requirements.

Notes:

• Results in the paper are generated using NVIDIA CUDA 11.6