Alternative Miniconda3 and Cellpose3 installation (NOT recommended)
Please ONLY follow these steps if the Miniforge3 and Cellpose installation in the last section did NOT work.

Here you see, how to install Miniconda as well as Cellpose on Windows. Installation on MacOS is shown below.
Miniconda3 installation:

You find the page with the Miniconda installation instructions used in the following video here.
 
Cellpose installation on Windows

For Cellpose, installation instructions are available on the Cellpose GitHub page

Some more info regarding potential issues can be found on the Cellpose page.

If you encounter further problems not discussed here please consult the image.sc forum for more help.
Commands for cellpose 3.1.1.2 installation (see video below):

the following commands have to be entered in the command prompt one after the other. Easiest is, if you just copy the lines and paste them into your command prompt window to avoid typos.

conda create --name cellpose3 python=3.10

conda activate cellpose3

In the following command, make sure you also use the version indication (==3.1.1.2) otherwise the newest version 4.x will be installed which is different and does not work with all exercises in this course

pip install cellpose[gui]==3.1.1.2

Now we test to start Cellpose

cellpose

If there is an error in the console window that points to Numpy version 2.x or multichannel images (up to 3 channels) cannot be opened in cellpose the following should solve that problem.

pip uninstall numpy

pip install numpy==1.26.4

You can directly start cellpose again or, to be on the save side, close the command prompt window and reopen it by pressing the Windows key, then type cmd and hit Enter.

conda activate cellpose3

cellpose

There should no error occur anymore only some startup info on cellpose.
 
Getting cellpose use the GPU on Windows (optional):

This part is not necessary doing the course and exercises but it might speed up calculation times. To make this really work you would also need to check that your graphic cards driver is up-to-date (not shown here).

pip uninstall torch

The following command will install cuda version 12.6 which might not work with all kinds of graphic cards. The video also shows where you find information on additional versions.

pip3 install torch torchvision --index-url https://download.pytorch.org/whl/cu126
 
Here you see, how to install the Miniconda as well as Cellpose on MacOS.

Disclaimer: We cannot guarantee that the installation on MacOS works exactly as shown in this video and we cannot provide trouble shooting in case it does not work. Some issue solving is already discussed during the video but for more detailed in the miniconda3 quick installation instructions.
 
Cellpose installation on MacOS (see video below)

For Cellpose, installation instructions are available on the Cellpose GitHub page

Some more info regarding potential issues can be found on the Cellpose page.

If you encounter further problems not discussed here please consult the image.sc forum for more help.
Commands for cellpose 3.1.1.2 installation on MacOS:

the following commands have to be entered in the command prompt one after the other. Easiest is, if you just copy the lines and paste them into your command prompt window to avoid typos.

conda create --name cellpose3 python=3.10

conda activate cellpose3

In the following command, make sure you also use the version indication (==3.1.1.2) otherwise the newest version 4.x will be installed which is different and does not work with all exercises in this course

pip install 'cellpose[gui]==3.1.1.2'


If this installation process stalls at some point when trying to install a package called opencv-python-headless and does run forever without finishing then close the terminal window, open a new one and try the following before continuing with the standard installation:

conda activate cellpose3

conda install py-opencv==4.10     OR     conda install -c conda-forge py-opencv==4.10

pip install 'cellpose[gui]==3.1.1.2'

cellpose


If there is an error pointing to a missing pyqt library/installation the following should solve that problem:  

conda install pyqt==5.15.10     OR     conda install -c conda-forge pyqt==5.15.10


If there are errors in the terminal window that points to Numpy version 2.x or multichannel images (up to 3 channels) cannot be opened in cellpose the following should solve that problem: 
 
pip uninstall numpy

pip install numpy==1.26.4


Close the command prompt window and reopen it by pressing the Windows key, then type cmd and hit Enter. 
 
conda activate cellpose3

cellpose

There should no error occur anymore only some startup info on cellpose.