Skip to content

Latest commit

 

History

History
57 lines (41 loc) · 2.03 KB

INSTALL.md

File metadata and controls

57 lines (41 loc) · 2.03 KB
Raw

Prerequisites

The code has been tested in the environment described as follows:

An example script for installing the python dependencies under CUDA 11.3:

# Export the PATH of CUDA toolkit
export PATH=/usr/local/cuda-11.3/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-11.3/lib64:$LD_LIBRARY_PATH

# Create conda environment
conda create -y -n epropnp_det python=3.7
conda activate epropnp_det
conda install -y pip

# Install pytorch
pip install torch==1.10.1+cu113 torchvision==0.11.2+cu113 torchaudio==0.10.1 -f https://download.pytorch.org/whl/torch_stable.html

# Install MMCV
pip install mmcv-full==1.4.1 -f https://download.openmmlab.com/mmcv/dist/cu113/torch1.10.0/index.html

# Install Pytorch3D dependencies
conda install -y -c fvcore -c iopath -c conda-forge -c bottler fvcore iopath nvidiacub

# Install Pytorch3D from source
git clone https://github.com/facebookresearch/pytorch3d
cd pytorch3d && git checkout v0.6.1 && pip install -v -e . && cd ..
# alternatively if you use pytorch 1.10.0, PyTorch3D can be directly installed via conda:
# conda install -y pytorch3d==0.6.1 -c pytorch3d

Installation

Clone the repository and install epropnp_det:

git clone https://github.com/tjiiv-cprg/EPro-PnP && cd EPro-PnP/EPro-PnP-det
pip install -v -e .

Verification

To verify the installation, you can download one of the checkpoint files [Google Drive | Baidu Pan] and run the inference demo:

python demo/infer_imgs.py demo/ /PATH/TO/CONFIG /PATH/TO/CHECKPOINT --show-views 3d bev mc

The resulting visualizations will be saved into demo/viz.