Galaxy Zoo DESI: Morphology Classifications for 8.7 Million Galaxies

Galaxy Zoo DESI provides detailed morphology classifications for 8.7 million bright galaxies from the DESI Legacy Imaging Surveys. The catalogs contain predicted vote fractions for features like bars, spiral arms, and mergers, generated by deep learning models trained on volunteer classifications. Two catalog versions are available, offering either a simplified or an advanced presentation of the morphology measurements and their uncertainties.

Use Cases

• [Analyzing the prevalence of galactic bars] based on predicted vote fractions for bar presence.
• [Studying spiral arm structure] based on classifications for spiral features and arm counts.
• [Identifying galaxy mergers] based on predicted vote fractions for merger activity.
• [Training or benchmarking new machine learning models] based on a large-scale, expert-validated morphology catalog.
• [Conducting statistical studies of galaxy populations] based on morphology measurements for millions of objects.

Strengths

• Contains morphology predictions for a large scale of 8.7 million galaxies.
• Leverages a proven annotation method via models trained on millions of Galaxy Zoo volunteer responses.
• Provides two catalog versions tailored for different user expertise levels, with the advanced version including estimated credible intervals.

Limitations

• Key metadata such as column names, row counts per file, and exact update dates are not provided in the available sources.
• The morphology measurements are model predictions, not direct human classifications, which may introduce specific biases.
• The dataset's cross-platform presence is limited to a single platform type (paperswithcode), suggesting potential discoverability gaps.

Provenance

Source

Data released from the paper 'Galaxy Zoo DESI: Detailed Morphology Classifications for 8.7M Galaxies in the DESI Legacy Imaging Surveys' by Mike Walmsley (University of Manchester).

Collection Method

Deep learning models (available via Zoobot) trained on Galaxy Zoo volunteer responses were used to predict morphology vote fractions for new galaxies.

Geography

Covers galaxies in the DESI Legacy Imaging Surveys footprint.