Dependencies

Napari-Easy-Augment-Batch-DL relies on one or more deep learning frameworks being installed in your environment. On start-up it attempts to import the following deep learning toolkits. At least one has to be successfully imported

Cellpose: For instance segmentation
Stardist: For instance segmentation
Pytorch: For semantic segmentation
Mobile SAM: An approach that uses Yolo + SAM for instance segmentation.

Other frameworks can be added via our plugin mechanism (Documentation to come)

Below are sets of suggested installation instructions for Linux, Mac M1, and Windows. Not all dependencies are needed. You need atleast one deep learning framework, but for example if you would like to work with Cellpose only you do not need Stardist or SAM.

Linux

    conda create -n easy_augment_env python=3.11
    conda activate easy_augment_env
    conda install pip
    pip install "napari[all]" # requires quotation marks to parse the square brackets on Linux, not sure if generalizes
    pip install albumentations matplotlib scipy tifffile 
    pip install "tensorflow[and-cuda]" # as above, requires quotation marks (only needed for stardist)
    pip install stardist 
    pip install gputools #==0.2.15 may no longer needed if we are happy with np 2
    pip install edt # pip throws: numba v0.60.0, tensorflow v2.18.0 require lower versions of numpy. Should still work but iffy.
    pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    pip install pytorch-lightning
    pip install git+https://github.com/Project-MONAI/MONAI # as of early spring 2025 need to get the github version of monai if using numpy 2.x 
    pip install cellpose
    pip install git+https://github.com/True-North-Intelligent-Algorithms/tnia-python.git 
    pip install git+https://github.com/True-North-Intelligent-Algorithms/segment-everything.git
    pip install git+https://github.com/True-North-Intelligent-Algorithms/napari-easy-augment-batch-dl.git
    pip install git+https://github.com/True-North-Intelligent-Algorithms/napari-segment-everything.git

Mac M1

    conda create -n easy_augment_env python=3.11
    conda activate easy_augment_env
    pip install "napari[all]" # also requires quotes on Mac
    pip install albumentations matplotlib scipy tifffile 
    pip install "tensorflow[and-cuda]" # as above, requires quotation marks. Pip throws a bunch of conflicts, but whatever. (only needed for stardist)
    pip install stardist
    pip install gputools # version may no longer be needed if going with numpy 2.x ==0.2.15 # pip dependency incompatibilities
    pip install edt 
    pip install torch torchvision torchaudio # remove the index flag url
    pip install pytorch-lightning
    pip install git+https://github.com/Project-MONAI/MONAI # as of early spring 2025 need to get the github version of monai if using numpy 2.x 
    pip install cellpose
    pip install git+https://github.com/True-North-Intelligent-Algorithms/segment-everything.git
    pip install git+https://github.com/True-North-Intelligent-Algorithms/napari-easy-augment-batch-dl.git
    pip install git+https://github.com/True-North-Intelligent-Algorithms/tnia-python.git 
    pip install git+https://github.com/True-North-Intelligent-Algorithms/napari-segment-everything.git

Windows

Note that to use GPU with Stardist in Windows we need two different environments because Windows needs an older version of tensorflow, which is not compatible wither newer version of pytorch based toolkits. This has led to increasingly convoluted installation instructions for stardist-GPU-windows (as transitive dependencies become an issue). See this thread. Please add to it if you run into issues.

I've also archived a windows gpu stardist environment.yml here. You will need both the environment.yml and requirements.txt.

Current Windows stardist GPU environment instructions ()

    conda create -n easy_augment_stardist_env python=3.11
    conda activate easy_augment_stardist_env
    pip install numpy==1.26.4 # start with numpy 1.26 and hope nothing upgrades it...
    pip install "napari[all]"
    pip install albumentations matplotlib scipy tifffile 
    conda install -c conda-forge cudatoolkit=11.8 cudnn=8.1.0
    pip install "tensorflow<2.11"
    pip install stardist==0.8.5
    pip install gputools==0.2.15
    pip install edt
    pip install reikna==0.8.0 
    pip install numpy==1.26.4 # in case numpy got upgraded go back (hacky yes, but this is what people are reporting works)
    pip install numba==0.59.1
    pip install git+https://github.com/True-North-Intelligent-Algorithms/tnia-python.git 
    pip install git+https://github.com/True-North-Intelligent-Algorithms/napari-easy-augment-batch-dl.git

Windows pytorch environment

    conda create -n easy_augment_pytorch_env python=3.11
    conda activate easy_augment_pytorch_env
    pip install "napari[all]"
    pip install albumentations matplotlib scipy tifffile 
    pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    pip install pytorch-lightning
    pip install git+https://github.com/Project-MONAI/MONAI # as of early spring 2025 need to get the github version of monai if using numpy 2.x 
    pip install cellpose
    pip install git+https://github.com/True-North-Intelligent-Algorithms/segment-everything.git
    pip install git+https://github.com/True-North-Intelligent-Algorithms/napari-easy-augment-batch-dl.git
    pip install git+https://github.com/True-North-Intelligent-Algorithms/tnia-python.git 
    pip install git+https://github.com/True-North-Intelligent-Algorithms/napari-segment-everything.git