INSTALL.md

Installation (choose to either install CUDA 10.1 (step 1a) or 11.1 (step 1b))

1a.) Install CUDA 10.1 (if not yet installed)

Tested on: Ubuntu 16.04 and 18.04, CUDA: 10.1 (10.1.105)

1a.1) Install NVIDIA drivers (using the terminal):

• sudo add-apt-repository ppa:graphics-drivers/ppa
• sudo apt update
• sudo apt install nvidia-driver-440
• nvidia-smi
• sudo reboot
  
  

1a.2) Download CUDA 10.1:

• download cuda 10.1: https://developer.nvidia.com/cuda-10.1-download-archive-base?target_os=Linux&target_arch=x86_64&target_distro=Ubuntu&target_version=1804&target_type=deblocal
• download the deb file, for example: cuda-repo-ubuntu1804-10-1-local-10.1.105-418.39_1.0-1_amd64.deb
  
  

1a.3) Install CUDA 10.1 (using the terminal, cd to the directory where the file has been downloaded):

• sudo dpkg -i cuda-repo-ubuntu1804-10-1-local-10.1.105-418.39_1.0-1_amd64.deb
• sudo apt-key add /var/cuda-repo-10-1-local-10.1.105-418.39/7fa2af80.pub
• sudo apt-get update
• sudo apt-get install cuda-toolkit-10-1
  
  

1a.4) Set the environmental path:

• sudo gedit ~/.bashrc
• add the following 2 lines at the end of the bashrc file: export PATH=/usr/local/cuda-10.1/bin${PATH:+:${PATH}} export LD_LIBRARY_PATH=/usr/local/cuda-10.1/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
• save the bashrc file and close it
• source ~/.bashrc
  
  

1a.5) Check if CUDA has been installed properly:

• nvcc --version (should the CUDA details)
  
  

1b.) Install CUDA 11.1.1 (if not yet installed)

Tested on: Ubuntu 20.04, and 22.04, CUDA: 11.1.1 (11.1.105-1)

1b.1) Install NVIDIA drivers (using the terminal):

• sudo add-apt-repository ppa:graphics-drivers/ppa
• sudo apt update
• sudo apt install nvidia-driver-460
• nvidia-smi
• sudo reboot
  
  

1b.2) Install CUDA 11.1.1:

• wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pin
• sudo mv cuda-ubuntu2004.pin /etc/apt/preferences.d/cuda-repository-pin-600
• wget https://developer.download.nvidia.com/compute/cuda/11.1.1/local_installers/cuda-repo-ubuntu2004-11-1-local_11.1.1-455.32.00-1_amd64.deb
• sudo dpkg -i cuda-repo-ubuntu2004-11-1-local_11.1.1-455.32.00-1_amd64.deb
• sudo apt-key add /var/cuda-repo-ubuntu2004-11-1-local/7fa2af80.pub
• sudo apt-get update
• sudo apt-get install cuda-toolkit-11-1
  
  

1b.3) Set the environmental path:

• sudo gedit ~/.bashrc
• add the following 2 lines at the end of the bashrc file: export PATH=/usr/local/cuda-11.1/bin${PATH:+:${PATH}} export LD_LIBRARY_PATH=/usr/local/cuda-11.1/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
• save the bashrc file and close it
• source ~/.bashrc
  
  

1b.4) Check if CUDA has been installed properly:

• nvcc --version (should the CUDA details)
  
  

2.) Install MaskAL in a virtual environment (using Anaconda)

Tested with: Pytorch 1.8.0 & torchvision 0.9.0 (Ubuntu 16.04/18.04 and CUDA 10.1), Pytorch 1.7.1 & torchvision 0.8.2 (Ubuntu 20.04 and CUDA 11.1), Pytorch 1.9.0 & torchvision 0.10.0 (Ubuntu 22.04 and CUDA 11.1)

2.1) Download and install Anaconda:

• download anaconda: https://www.anaconda.com/distribution/#download-section (python 3.x version)
• install anaconda (using the terminal, cd to the directory where the file has been downloaded): bash Anaconda3-2019.10-Linux-x86_64.sh
  
  

2.2) Make a virtual environment (called maskal) using the terminal:

• conda create --name maskal python=3.9 pip
• conda activate maskal
  
  

2.3) Downgrade setuptools, to prevent this error:

• pip uninstall setuptools
• pip install setuptools==59.5.0
  
  

2.4) Download the code repository:

• git clone https://github.com/pieterblok/maskal.git
• cd maskal
  
  

2.5) Install the required software libraries (in the maskal virtual environment, using the terminal):

• for cuda 10.1: pip install -U torch==1.8.0 torchvision==0.9.0 -f https://download.pytorch.org/whl/cu101/torch_stable.html
• for cuda 11.1: pip install -U torch==1.7.1 torchvision==0.8.2 -f https://download.pytorch.org/whl/cu111/torch_stable.html
• for cuda 11.1: pip install -U torch==1.9.0 torchvision==0.10.0 -f https://download.pytorch.org/whl/cu111/torch_stable.html
• pip install pillow==9.0.1
• pip install cython
• pip install jupyter
• pip install opencv-python
• pip install -U fvcore
• pip install scikit-image matplotlib imageio
• pip install black isort flake8 flake8-bugbear flake8-comprehensions
• pip install -e .
• pip install baal
• pip install xmltodict
• pip install seaborn
• pip install statsmodels
• pip install cerberus
• pip install darwin-py
  
  

2.6) Reboot/restart the computer (sudo reboot)

2.7) Check if Pytorch links with CUDA (in the maskal virtual environment, using the terminal):

• python
• import torch
• torch.version.cuda (should print 10.1 or 11.1)
• torch.cuda.is_available() (should True)
• torch.cuda.get_device_name(0) (should print the name of the first GPU)
• quit()
  
  

2.8) Check if detectron2 is found in python (in the maskal virtual environment, using the terminal):

• python
• import detectron2 (should not print an error)
• from detectron2 import model_zoo (should not print an error)
• quit()