First contact with deepImageJ 2.1

If it is the first time that you are using deepImageJ or you've just installed the latest version of it, we recommend you the following super easy steps:
Try a model direclty in ImageJ
  • Click on DeepImageJ Install, and install one of the suggested models compatible with deepImageJ from the list. This list refers to all the models in the BioImage Model Zoo. The ones compatible whit deepImageJ are marked in green. There's no need to restart Fiji once the model is installed.
  • Once the model is installed, you can find a demo image (tif) in the folder created at Fiji/models.
  • Open the image in ImageJ and process it with deepImageJ (DeepImageJ Run).
Try a model from the BioImage Model Zoo
Do you want to create your own model package compatible with the BioImage Model Zoo?

If you are familiarized with deep learning models and architectures, this shouldn't be too difficult.

  • Download a deepImageJ model from the BioImage Model Zoo. You will (re)build the bundled model yourself with DeepImageJ Build Bundled Model.
  • This step assumes that you are using a "raw Python model".
  • Go to the zip file you just downloaded and extract the zip called "tensorflow_saved_model_bundle.zip". This will result in a "saved_model.pb" and a "variables" folder.
  • Run DeepImageJ Build Bundled Model.
  • Drag&drop the folder containing saved_model.pb and a variables folder (it doesn't matter if there are other files).
  • Follow the steps of the user guide. As you are not the model creator you may need to check some parameters in the file called model.yaml in the zip file that you downloaded.
  • The pre- and post-processing files and example data are also in that zip file.
DeepImageJ Run is macro recordable!
  • Start the macro recorder.
  • Use DeepImageJ Run on an image.
  • Copy the code line given by the macro recorder.
  • Create a new ImageJ macro, paste the line and run it.

Tutorials and seminars about image processing with deep learning and deepImageJ 1

Practical introduction to deepImageJ 1

Workshop at SPAOM 2020: Practical Applications of Deep learning for Bioimage Analysis: Slides

User guide tutorial

NEUBIAS Academy @Home

Webinar Intro to Machine Learning-DeepLearning-DeepimageJ by Ignacio Arganda-Carreras

Ecole Polytechnique Fédérale de Lausanne
Initiative Imaging@EPFL
Biomedical Imaging Group

GitHub repository

Universidad Carlos III de Madrid
Bioengineering and Aerospace Engineering Department