Zoobot Documentation

Zoobot makes it easy to finetune a state-of-the-art deep learning classifier to solve your galaxy morphology problem. For example, you can finetune a classifier to find ring galaxies with just a few hundred examples.

Ring galaxies found using Zoobot and 212 labelled examples.

The easiest way to learn to use Zoobot is simply to use Zoobot. We suggest you start with our worked examples.

• This Colab notebook will walk you through using Zoobot to classify galaxy images.

• There’s a similar notebook for using Zoobot for regression on galaxy images.

For more explanation, read on.

User Guides

These introductory guides add context to the demo Colab notebooks.

• Finetuning
• Choosing Parameters
• Loading Data
• Training on Vote Counts

These advanced guides explain how to integrate Zoobot into other ML projects.

• Advanced Finetuning
  • Using Zoobot’s Encoder Directly
  • Extracting Frozen Representations
  • Subclassing FinetuneableZoobotAbstract
• How the Code Fits Together
  • The Map
  • Finetuning with Zoobot Classes
  • Finetuning with PyTorch Lightning

Pretrained Models

To choose and download a pretrained model, see here.

• Pretrained Models
  • Loading Models
  • Available Models
  • Which model should I use?
  • How were the models trained?

Science-Ready Data

You can find our science outputs (e.g. morphology catalogs, precalculated representations) here.

• Science Data
  • Precalulated Representations
  • Galaxy Zoo Morphology

We are working on releasing the compiled GZ Evo dataset and will update this page when it is available. Estimated public release is Q4 2024. Please reach out if you’d like early access.

API reference

We’ve added docstrings to all the key methods you might use. Feel free to check the code or reach out if you have questions.

• pytorch
  • estimators
    • define_model
      • ZoobotTree
      • get_pytorch_encoder()
      • get_pytorch_dirichlet_head()
    • efficientnet_custom
      • custom_top_dirichlet()
  • training
    • train_with_pytorch_lightning
      • train_default_zoobot_from_scratch()
    • finetune
      • FinetuneableZoobotAbstract
      • FinetuneableZoobotClassifier
      • FinetuneableZoobotRegressor
      • FinetuneableZoobotTree
      • LinearHead
      • load_pretrained_zoobot()
      • get_trainer()
      • download_from_name()
    • losses
      • calculate_multiquestion_loss()
      • dirichlet_loss()
  • predictions
    • predict_on_catalog
      • predict()

• shared
  • schemas
    • Question
    • Answer
    • create_answers()
    • set_dependencies()
    • Schema