Creating Simulation environment

• Creating Simulation environment
  • Setup virtual environment (Not compulsory)
  • Install Unreal Engine 4.22
  • Install Git Large File Storage (LFS)
  • Install NVIDIA Deep learning Dataset Synthesizer (NDDS)
  • Blender
  • Labelme
• Tutorials

Setup virtual environment (Not compulsory)

• Open the terminal and run the following commands

  ​​​​mkdir 3d && cd 3d
  ​​​​virtualenv venv
  ​​​​source venv/bin/activate
  

Install Unreal Engine 4.22

• Follow the instructions written on the official website, or follow the steps below.
  

• Register to an Epic Games Account and connect it with your github account.

• Accept the invitation of Epic Games on Github.

• In the terminal run the following comands,

  ​​​​git clone -b 4.22 https://github.com/EpicGames/UnrealEngine.git
  

  
  Enter your github login information

  ​​​​cd UnrealEngine  # 1.4 GB
  ​​​​./Setup.sh
  ​​​​./GenerateProjectFiles.sh
  ​​​​make
  

  • Select Yes, if the below window pops-up
    

• Setting UnrealEngine on .bashrc

  • Open .bashrc

    ​​​​​​​​nano ~/.bashrc
    

  • Add this lines to .bashrc file

    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    ​​​​​​​​UE_HOME=/home/jitesh/3d/UnrealEngine
    
    ​​​​​​​​function unrealbuild {
    ​​​​​​​​         CURR_DIR=`pwd`;
    ​​​​​​​​         PROJ_NAME=$(basename ${1%.uproject});
    ​​​​​​​​         $UE_HOME/Engine/Build/BatchFiles/Linux/Build.sh $PROJ_NAME Linux Development -editorrecompile "${CURR_DIR}/${PROJ_NAME}.uproject" -progress -editor -game -NoHotReloadFromIDE
    ​​​​​​​​}
    ​​​​​​​​complete -f -X '!*.@(uproject)' unrealbuild
    
    ​​​​​​​​function unrealeditor {
    ​​​​​​​​    DIR="$( cd "$( dirname "$i" )" && pwd )"
    ​​​​​​​​    $UE_HOME/Engine/Binaries/Linux/UE4Editor $DIR/$1
    ​​​​​​​​}
    ​​​​​​​​complete -f -X '!*.@(uproject)' unrealeditor
    
    ​​​​​​​​function unrealgen {
    ​​​​​​​​    DIR="$( cd "$( dirname "$i" )" && pwd )"
    ​​​​​​​​    pushd $UE_HOME
    ​​​​​​​​    ./GenerateProjectFiles.sh -project="$DIR/$1" -game -engine -editor
    ​​​​​​​​    popd
    ​​​​​​​​}
    ​​​​​​​​complete -f -X '!*.@(uproject)' unrealgen
    
    ​​​​​​​​# For automation tests, if used
    ​​​​​​​​function unrealtest {
    ​​​​​​​​    DIR="$( cd "$( dirname "$i" )" && pwd )"
    ​​​​​​​​        $UE_HOME/Engine/Binaries/Linux/UE4Editor $DIR/$1 -Game -ExecCmds="Automation RunTests $2" -log
    ​​​​​​​​}
    ​​​​​​​​complete -f -X '!*.@(uproject)' unrealtest
    

  • Change UE_HOME to the path of your UnrealEngine directory

    ​​​​​​​​sudo sh ~/.bashrc
    

• Reboot the PC

• Running UE4

  ​​​​cd Engine/Binaries/Linux/
  ​​​​./UE4Editor
  

Install Git Large File Storage (LFS)

• Click here to download the package or download it from the lfs site.
• Uncompress the package and run the following command from inside the package.

  ​​​​sudo ./install.sh
  ​​​​git lfs install
  

Install NVIDIA Deep learning Dataset Synthesizer (NDDS)

• Download NDDS using 'Method 1' or 'Method 2'
  • Method 1
    Enter the following commands in your workspace

    ​​​​​​​​git lfs clone https://github.com/NVIDIA/Dataset_Synthesizer.git
    

  • Method 2
    If the 'git lfs clone' doen't work, then download NDDS using this link.
• Running the NDDS.uproject

  ​​​​cd Dataset_Synthesizer/Source
  ​​​​unrealgen NDDS.uproject
  ​​​​unrealbuild NDDS.uproject
  ​​​​unrealeditor NDDS.uproject
  

  • Select Yes, if the below window pops-up
    
  • When it starts, it should look like this
    
    
    
    
    

Blender

• sudo snap install blender --classic

Labelme

• Install

  ​​​​sudo apt-get install python3-pyqt5  -y
  ​​​​pip3 install labelme
  

• Run

  ​​​​labelme --nodata --autosave -o json img
  

Tutorials

• For NDDS, follow "Dataset_Synthesizer/Documentation/NDDS.pdf"