Installation instructions

← Back to main page

Windows

System requirements:

  • 64-bit Windows, version 7 and later

  • Video card with CUDA 11 support and 2 GB of video RAM

Additional system requirements for training the neural network:

  • Video card with CUDA 11 support and 11 GB of video RAM

  • 8 GB RAM if you want to train without using the intermediary TFRecord files

First, make sure you have Anaconda or Miniconda installed. You can download them at anaconda.com/distribution. You should install the Anaconda version that ships with Python 3, although the one that ships with Python 2 will also work (Anaconda will just download Python 3 afterwards.) Using the default installation options of Anaconda is fine.

Second, make sure you have downloaded OrganoidTracker. You can install it through Git (if you’re familiar with that program), or simply download it from here and then extract it somewhere.

Next, to run the scripts you will first need to install the dependencies. Open the Anaconda Prompt (in the “All Programs” list of your start menu, there should now be a folder named Anaconda3, which contains the Anaconda Prompt). Use the command cd path\to\folder\with\OrganoidTracker (replace path with real path) to navigate to the directory you installed OrganoidTracker in. If you are not used to the command line, you can also type “cd ” (a c, a d and a space) and then drag in the OrganoidTracker folder and drop it in the command prompt, and then press Enter. See this illustration:

Dragging and dropping a folder

Once you’re in the right directory, run the following two commands.

conda env create -f environment-exact-win64.yml
conda activate organoid_tracker

The last command is necessary because Tensorflow 2 is not available in Anaconda. If you need to remove (the previous version of) OrganoidTracker, execute this command:

conda env remove -n organoid_tracker

To test if the software is working, run python organoid_tracker.py. A window should pop up, from which you can load images and tracking data. See the Help menu for more information and tutorials.

Note: in previous versions of OrganoidTracker you would need to install CUDA and cuDNN. Because those programs are now installed via Anaconda, you no longer need to do this.

Tip: you can install the orjson library into your Conda environment to speed up saving/loading of AUT files by a factor of two. To do this, run the command conda install --channel conda-forge orjson.

macOS and Linux

Unfortunately, OrganoidTracker has not been tested on any other OSes. If you’re in for a challenge, here are some instructions. Feel free to ask the authors if you run into any problems, we’re interested in getting OrganoidTracker to work on other OSes.

Download and install Anaconda/Miniconda and open the Anaconda Prompt. Use the cd command to navigate to the directory of OrganoidTracker. Run the following commands:

conda create -n organoid_tracker -c ilastik-forge -c esri -c conda-forge python=3.9 dpct matplotlib-base tifffile scikit-image scipy nd2reader pyside2 tensorflow-gpu tensorflow-addons
source activate organoid_tracker

The first command creates an Anaconda environment named “organoid_tracker” with all dependencies installed. The second command activates this environment. The third command installs Tensorflow, which only has a PIP package.

If you need to remove (the previous version of) OrganoidTracker, execute this command:

conda env remove -n organoid_tracker

To test if the software is working, run python organoid_tracker.py. A window should pop up, from which you can load images and tracking data.