Installation

This guide walks you through installing Napari-OmniEM from a downloaded archive and preparing the runtime environment.

Step 1: Download Napari-OmniEM

Download the napari-omniem.zip archive from the given google driver folder or temporarily open link and extract it to a local directory:

unzip napari-omniem.zip
cd napari-omniem

Step 2: Create a Conda Environment

We recommend using conda to manage dependencies.

conda create -n napari-omniem python=3.11 -y
conda activate napari-omniem

Step 3: Install PyTorch with CUDA Support

Install PyTorch with the appropriate CUDA version for your system.

# Example: CUDA 11.8
pip install torch torchvision torchaudio \
  --index-url https://download.pytorch.org/whl/cu118

Visit PyTorch to find the correct installation command for your GPU and CUDA version. If you do not have a GPU, install the CPU-only version instead.

Step 4: Install Napari

Install napari and its GUI dependencies

pip install napari[all]

Step 5: Install Napari-OmniEM

Install the plugin in editable (development) mode:

pip install -e .

This makes the plugin immediately available in napari.

Step 6: Launch Napari

Start napari from the command line:

napari

Once napari is running, you should be able to find OmniEM in the plugin menu.

Location of OmniEM in Plugins

Notes

  • Ensure that your GPU drivers and CUDA runtime are properly installed before installing PyTorch with CUDA support.
  • For large-volume inference, sufficient GPU memory and disk space are required.
  • Multi-GPU inference is supported when multiple CUDA devices are available.