# Revisiting the SFR-Mass relation at z=0 with detailed deep learning based morphologies

## TL;DR

This paper presents a large deep learning-based morphological catalogue for 670,000 SDSS galaxies, enabling detailed analysis of galaxy morphology's impact on the star formation-stellar mass relation.

## Contribution

It introduces the largest T-Type galaxy classification catalogue derived from deep learning on SDSS images, with uncertainties and multiple morphological labels.

## Key findings

- The SFR-M* relation varies with galaxy morphology.
- Deep learning effectively classifies galaxy morphologies at large scale.
- The catalogue enables new insights into galaxy evolution studies.

## Abstract

Galaxy morphology is a key parameter in galaxy evolution studies. The enormous number of galaxies which current and future surveys will observe demand of automated methods for morphological classification. Supervised learning techniques have been successfully used for the morphological classification of galaxies from different datasets, including Sloan Digital Sky Survey (SDSS), Mapping Galaxies with Apache Point Observatory (MaNGA) or Dark Energy Survey (DES). With these proceedings, we release the morphological catalogue for a sample of 670,000 SDSS galaxies based on the deep learning models trained on SDSS RGB images with morphological labels from human-based classification catalogues. The released catalogue includes binary classifications (early-type versus late-type, elliptical versus lenticular, identification of edge-on and barred galaxies) plus a T-Type. The classifications also include k-fold based uncertainties. This is, as of today, the largest catalogue including a T-Type classification. As an example of the scientific potential of this classification, we show how the location of the galaxies in the star formation - stellar mass plane (SFR-M$^{*}$) depends on morphology. This is the first time the SFR-M$^{*}$ relation is combined with T-Type information for such a large sample of galaxies.