Download: Shared OneDrive Folder
Inside you will find a Data
folder and an ExternalPackages
folder containing the cudnn and tensorflow packages. Copy both folders to the root of the repository. You will need them later...
NOTE: This step is optional for hardware acceleration IF you have a compatible NVidia card. If not, or for slow workflow but easier install, skip this bit.
(Full instructions, but we shouldn't need this!)
In order to use the graphics card for hardware accelerated inference in Bonsai, you will need to install the Cuda Toolkit 11.3 (for SLEAP multi-animal tracking)
- Select Custom install and check
CUDA > Development
andCUDA > Runtime
ONLY (uncheck everything else)
Once installed, could be worth checking your User and System Variables. You should see your Python 3.10 installation on the path of User Variables. You should see the CUDA_PATH is correctly pointing to CUDA 11.3, and the Cuda/bin folder is on the system variables path.
- In a command terminal, run
setup.cmd
from the.bonsai
folder . This will run the powershell scriptsetup.ps1
to automatically intall a local Bonsai installation you can use to edit and run the Workflows in this project
Direct Download: python 3.10 (Windows 64-bit installer) Python release page
Hit Ctrl-P to bring up the command pallette. Choose Python: Create Environment...
, then Venv
.
Choose your Python 3.10 installation as the interpreter path, and check the requirements.txt
.
Done. If you now open a new Powershell terminal, the environment will activate automatically.
Install virtual env if needed
pip install virtualenv
Create the virtual environment
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
.\.bonsai\Bonsai.exe
git clone https://github.com/jfrazao/BonFly.git
Install Bonsai by running the setup.cmd
in the .\bonsai
folder
- Download and run Visual Studio Community.
- Check .NET Desktop Development workload
- (optional) uncheck IntelliCode support.
From the Windows Start Menu, search for the "Install Bonsai VS Extensions" shortcut and run it.
- Download and install Microsoft Child Process Debugging Power Tool.
- For each project:
- Ensure that
launchSettings.json
is set to"nativeDebugging": true
. - Enable child process debugging in the
Debug
>Other Debug Targets
>Child Process Debugging Settings
.
- Ensure that
Navigate to TOOLS->Options->Debugging->General
and disable "Require source files to exactly match the original version" .
Go to TOOLS->Options->Debugging->Symbols
and enable the Microsoft Symbols Server.
Finally, under TOOLS->Options->Debugging
, check "Enable Just My Code' and "use Manage/Unmanage mode".
- Download and run Visual Studio Code.
- Make sure "Add to PATH" option is checked.
- Download and run Git for Windows.
- Select components: uncheck all except Git LFS.
- Default editor: Visual Studio Code.
- Override default branch name: main
- Choose "Git from the command line and also from 3rd party software".
- Use bundled OpenSSH.
- Use native Windows Secure Channel library.
- Checkout Windows-style, commit Unix-style line endings.
- Use default console window.
- Default (fast-forward or merge).
- Use default Git credential manager.
- Leave at defaults for everything else.
- Open command prompt (cmd.exe) or PowerShell.
- Configure git user and email