1. Open Docker Desktop.
2. Download a ZIP file from: https://downloads.zeiss.com/arivis/Pro/latest/cellpose-onnx (<100KB).
3. Extract the ZIP.
4. Double click on run_cellpose-onnx.bat file.
5. It might take some time as all the needed components get downloaded and composed into a Docker container image.
6. The container gets started automatically.
5. Your web browser should open at http://127.0.0.1:5000/ automatically.
6. After starting the tool you can decide if you want to convert models compatible with Cellpose 3 or Cellpose 4 by simply clicking in the UI.
   
7. At the first start the list of the available built-inCellpose models is downloaded.
8. Select one of the Cellpose model that you would like to download and convert.
9. The batch size can be adjusted if required; however, in most cases, the default value is recommended and does not need to be changed.
10. Review and confirm the applicable license terms carefully - for the build-in Cellpose models they can be found at https://github.com/mouseland/cellpose.
11. Click Download and Convert to ONNX to start the download and conversion process.
12. A progress indicator will be displayed below. Once the conversion has completed successfully, a confirmation message will appear along with the option to download the converted ONNX model to your local machine.
    
13. Every successfully converted model is listed in the Converted Files section and can be downloaded at any time. Save the converted model files to a suitable location on the same workstation as arivis Pro.
14. If the tool is running on the same workstation as arivis Pro a download is not needed as the converted models can be used directly from the configured output directory. By default, the models are stored in the output folder within the downloaded conversion tool package.
15. Best practice: Download and save the models into the Cellpose Models folder (usually located under Documents\Cellpose Models) to easily access them in the Cellpose-based segmentation operation. The folder can be adjusted via Analysis -> Cellpose preferences.

Changing the default output folder

You can change the default output folder by editing the `.env' file within the downloaded conversion tool package

1. Change the output path to the same path as the Cellpose models folder as configured in arivis Pro.
2. Use Python syntax (forward slashes), e.g.OUTPUT_PATH="C:/Users/arivis/Documents/Cellpose Models”
3. After change double click on run_cellpose-onnx.bat file again to rebuild the tool and apply the changes.
4. That way every converted model will be conveniently available to pick up in the Cellpose-based segmentation operation.
   

Converting custom models

Optional it is also possible to convert a custom model trained in Cellpose.

1. Upload a custom .pt model to convert.
   
2. To stop the converter tool, stop the container in Docker Desktop or quit Docker Desktop completely.
3. If you need the converter tool later again → run Docker Desktop and start the previously created container again.
4. Alternatively, double click on the run_cellpose-onnx.bat file again. This will start the container or recreate the container (e.g. if any update is available).