Installation

It is highly recomended to create and use a new environment to install the plugin and its dependencies. You can use mamba or conda to manage you environments but it's not necessary and you can use python venv as well.

Setup

We provided install.sh for Linux & Mac OS users, and install.bat for Windows users.
First you need to clone the repo:

git clone https://github.com/juglab/featureforest
cd ./featureforest

Now run the installation script:

# Linux or Mac OS
sh ./install.sh

# Windows
./install.bat

Step by Step

1. Create a new environment:

   conda create -n featureforest -y python=3.10
   

2. Activate the environment:

   conda activate featureforest
   

3. Install torch and torchvision:
   You can follow the instruction from here. But we can also use light-the-torch package:

   pip install light-the-torch
   ltt install 'torch>=2.5.1' 'torchvision>=0.20.1'
   

   This will install the appropriate PyTorch binaries without user intervention by automatically identifying compatible CUDA versions from the local setup.

4. Installing all other dependencies:

   pip install -r ./requirements.txt
   

   This will install all dependencies including napari, segment-anything and sam-2.

5. Finally, install the plugin:

   pip install git+https://github.com/juglab/featureforest.git
   

Requirements

• python>=3.10
• numpy<2.2
• pytorch>=2.5.1
• torchvision>=0.20.1
• timm
• segment-anything
• sam-2
• opencv-python
• scikit-learn
• scikit-image
• scipy
• matplotlib
• pyqt
• magicgui
• qtpy
• napari
• h5py
• pynrrd
• pooch

There is also a pypi package available that you can install FF using pip:

pip install featureforest

Note

Before install featureforest using pip you need to install segment-anything and sam-2 manually.

For detailed napari installation see here.