This repository will not be maintained in the future.
*** Please refer to https://github.com/nmsl-nthu/PCCArena for the latest version. ***
Ubuntu 20.04
- git
- gcc
- g++
- cmake
- subversion
- xvfb
- libblas-dev
- libatlas-base-dev
- nvidia-cuda-toolkit
sudo apt install git gcc g++ cmake subversion xvfb libblas-dev libatlas-base-dev nvidia-cuda-toolkit -y
- Ananconda 3
wget https://repo.anaconda.com/archive/Anaconda3-2020.11-Linux-x86_64.sh
sh Anaconda3-2020.11-Linux-x86_64.sh
source .bashrc
To download and set up PCC Arena, please type the following commands.
- Step 1: Clone the github project.
git clone https://github.com/xtorker/PCCArena.git
- Step 2: Change the current directory.
cd PCCArena
- Step 3: Create the conda environment pcc_arena.
conda env create -f cfgs/conda_env/pcc_arena.yml
- Step 4: Activate the environment pcc_arena.
conda activate pcc_arena
- Step 5: Set up the environments.
python setup.py
- Step 6: Grant executed permission.
chmod +x setup_env_ds.sh
- Step 7: Run environment setup script.
./setup_env_ds.sh
-
Step 8: Download the pretrained models of GeoCNNv1 and GeoCNNv2 using the following links. Save them in the current directory. GeoCNNv1: https://drive.google.com/file/d/1ayWqNHwvihAludemIF2l9NZSi6ofWHYy/view?usp=sharing GeoCNNv2: https://drive.google.com/file/d/1w5jue_dgR8Xw3D5gvZV1lXDA36NO5T_4/view?usp=sharing
-
Step 9: Uncompress the models into algorithms/GeoCNNv1 and algorithms/GeoCNNv2.
tar xvf geocnn_v1_pretrained_models.tar -C algorithms/GeoCNNv1
tar -Jxvf geocnn_v2_pretrained_models.tar.xz -C algorithms/GeoCNNv2
- Step 10: Run experiments in PCC Arena. We have two types of python files for experimenting. One is a short version for testing, and the other is a full version. The short version only runs one compression rate for each algorithm and doesn't run the algorithms which require lots of memory (e.g., GeoCNNv1 requires more than 50GB).
# Short version
python run_experiments_short.py
# Full version
python run_experiments.py
- Step 11: Check the results (binaries, point cloud, metrics) in
expereiments/{algorithm}/{rate}
- Put the whole PCC algorithm project folder under algorithms/
- Write a specific wrapper for it and put it under algs_wrapper/
- Write a YAML file for configuring any coding parameters and rate control parameters, and put it under cfgs/algs/
- (Optional) If the PCC algorithm needs specific virtual environment, make sure to indicate the python path in the YAML file (Step 3).