Outline

• Conda Setup
• RWTH Aachen CLAIX Cluster Setup
• Docker Setup
  • Build Docker Image
  • Run Docker Container
  • Install Chamfer Distance

Conda Setup

1. Create a conda virtual environment and activate it.

conda create -n occuq python==3.8 -y
conda activate occuq

2. Install netcal.

pip install netcal

3. Install PyTorch and torchvision (tested on torch==1.10.1 & cuda=11.3).

conda install pytorch==1.10.1 torchvision==0.11.2 torchaudio==0.10.1 cudatoolkit=11.3 -c pytorch -c conda-forge
conda install mkl==2024.0

4. Install gcc>=5 in conda env.

conda install -c omgarcia gcc-6 # gcc-6.2

5. Install MMCV following the official instructions.

pip install mmcv-full==1.4.0

6. Install mmdet and mmseg.

pip install mmdet==2.14.0
pip install mmsegmentation==0.14.1

7. Install ninja.

pip install ninja

8. Install mmdet3d from source code.

git clone https://github.com/open-mmlab/mmdetection3d.git
cd mmdetection3d
git checkout v0.17.1 # Other versions may not be compatible.
python setup.py install

9. Install other dependencies.

pip install scikit_image==0.19.3
pip install lyft-dataset-sdk==0.0.8 numba==0.48.0 nuscenes-devkit==1.1.10 plyfile==0.8.1 networkx==2.2 numpy==1.21.5
pip install timm
pip install open3d-python

10. Install Chamfer Distance.

cd OCCUQ/extensions/chamfer_dist
python setup.py install --user

11. Install other dependencies.

pip install yapf==0.40.1 setuptools==59.5.0
pip install einops --no-deps

RWTH Aachen CLAIX Cluster Setup

1. ssh onto the CLAIX GPU Node.

ssh username@login23-g-1.hpc.itc.rwth-aachen.de

2. Switch to GCC-based compiler.

module switch intel foss

2. Load the CUDA 11.3 module.

module load CUDA/11.3

3. Set environment variables.

export CC=gcc
export CXX=g++
export CUDAHOSTCXX=$(which g++)

4. Access node with V100 GPU.

srun --partition=dgx2 --nodes=1 --mem=512G --time=05:00:00 --gpus-per-node=1 --account=supp0003 --pty /bin/bash

Docker Setup

Build Docker Image

./docker/build.sh

Run Docker Container

./docker/run.sh

Attach to running container either with VS Code or via terminal.

docker exec -it occuq bash

Install Chamfer Distance

Inside the container execute the following commands to install the Chamfer Distance extension.

./docker/run_in_container.sh