Installation guide ⚙#

This guide outlines the steps for installing CellSeg3D and its dependencies. The plugin is compatible with Windows, Linux, and MacOS.

Note for ARM64 Mac Users: Please refer to the section below for specific instructions.

Warning

If you encounter any issues during installation, feel free to open an issue on our GitHub repository.

Installing pre-requisites#

PyQt5 or PySide2#

CellSeg3D requires either PyQt5 or PySide2 as a Qt backend for napari. If you don’t have a Qt backend installed:

pip install napari[all]

This command installs PyQt5 by default.

PyTorch#

For PyTorch installation, refer to PyTorch’s website , with or without CUDA according to your hardware. Select the installation criteria that match your OS and hardware (GPU or CPU).

Note

While a CUDA-capable GPU is not mandatory, it is highly recommended for both training and inference.

  • Running into MONAI-related errors? Consult MONAI’s optional dependencies for solutions. Please see MONAI’s optional dependencies page for instructions on getting the readers required by your images.

Installing CellSeg3D#

Warning

For ARM64 Mac users, please see the section below

Via pip:

pip install napari-cellseg3d

Directly in napari:

  • Navigate to Plugins > Install/Uninstall Packages

  • Search for napari-cellseg3d

For local installation (after cloning from GitHub) Navigate to the cloned CellSeg3D folder and run:

pip install -e .

Successful installation will add the napari-cellseg3D plugin to napari’s Plugins section.

ARM64 Mac installation#

For ARM64 Macs, we recommend using our custom CONDA environment. This is particularly important for ARM64 (Silicon chips) MacBooks.

Start by installing miniconda3.

Creating the environment#

  1. Clone the repository (link):

git clone https://github.com/AdaptiveMotorControlLab/CellSeg3D.git

2. Create the Conda Environment : In the terminal, navigate to the CellSeg3D folder:

cd CellSeg3D
conda env create -f conda/napari_cellseg3d_ARM64.yml

This will also install the necessary dependencies as well as the plugin.

  1. Activate the environment :

conda activate napari_cellseg3d_ARM64

4. Install a Qt backend : Important : you only need to install one of the following backends. PyQt5:

pip install PyQt5

OR PySide2:

pip install PySide2

5. Install PyTorch : Refer to PyTorch’s website for installation instructions.

6. Launch napari : You should now see the CellSeg3D plugin in the Plugins section of napari. See Usage section for a guide on how to use the plugin.

Updating the environment#

In order to update the environment, navigate to the CellSeg3D folder and run:

conda deactivate
conda env update -f conda/napari_cellseg3d_ARM64.yml

Troubleshoting#

pyClesperanto#

If you encounter the following error : clGetPlatformIDs failed: PLATFORM_NOT_FOUND_KHR :

Please install clinfo and check if your OpenCL platform is available.

If not, please install the OpenCL driver for your hardware.

[Source]

—

Please help us make this section better by reporting any issues you encounter during installation.

Optional requirements#

Additional functionalities#

Several additional functionalities are available optionally. To install them, use the following commands:

  • CRF post-processing:

pip install pydensecrf@git+https://github.com/lucasb-eyer/pydensecrf.git#egg=master

  • Weights & Biases integration:

pip install napari-cellseg3D[wandb]

  • ONNX model support (EXPERIMENTAL): Depending on your hardware, you can install the CPU or GPU version of ONNX.

pip install napari-cellseg3D[onnx-cpu]
pip install napari-cellseg3D[onnx-gpu]

Development requirements#

  • Building the documentation:

pip install napari-cellseg3D[docs]

  • Running tests locally:

pip install pydensecrf@git+https://github.com/lucasb-eyer/pydensecrf.git#egg=master
pip install napari-cellseg3D[test]

  • Dev utilities:

pip install napari-cellseg3D[dev]