Galaxies OBserved as Low-luminosity Identified Nebulae (GOBLIN): Catalog of 43 000 high-probability dwarf galaxy candidates in the UNIONS survey

¹ Institute of Physics, Laboratory of Astrophysics, Ecole Polytechnique Fédérale de Lausanne (EPFL), 1290 Sauverny, Switzerland
² Institute of Astronomy, Madingley Road, Cambridge CB3 0HA, UK
³ Visiting Fellow, Clare Hall, University of Cambridge, Cambridge, UK
⁴ National Research Council of Canada, Herzberg Astronomy & Astrophysics Research Centre, 5071 West Saanich Road, Victoria BC V9E 2E7, Canada
⁵ Department of Physics and Astronomy, University of Waterloo, 200 University Avenue West, Waterloo ON N2L 3G1, Canada
⁶ Institute for Astronomy, University of Hawaii, 2680 Woodlawn Drive, Honolulu, HI 96822, USA
⁷ Waterloo Centre for Astrophysics, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada
⁸ Perimeter Institute for Theoretical Physics, 31 Caroline St. North, Waterloo, ON N2L 2Y5, Canada
⁹ LIRA, Observatoire de Paris, Universite PSL, CNRS, Place Jules Janssen, 92195 Meudon, France
¹⁰ UK Astronomy Technology Centre, Royal Observatory, Blackford Hill, Edinburgh EH9 3HJ, UK

^⋆ Corresponding author: nick.heesters@epfl.ch

Received: 12 March 2025
Accepted: 23 May 2025

Abstract

The detection of low-surface-brightness galaxies beyond the Local Group poses significant observational challenges, yet these faint systems are fundamental to our understanding of dark matter, hierarchical galaxy formation, and cosmic structure. Their abundance and distribution provide crucial tests for cosmological models, particularly regarding the small-scale predictions of ΛCDM. We present a systematic detection and classification framework for unresolved dwarf galaxy candidates in the large-scale Ultraviolet Near Infrared Optical Northern Survey (UNIONS) imaging data. The main survey region covers 4861 deg². Our pipeline preprocesses UNIONS data in three (gri) of the five bands (ugriz), including binning, artifact removal, and stellar masking before employing the software MTOBJECTS (MTO) to detect low-surface-brightness objects. Following a set of parameter cuts using known dwarf galaxies from the literature and cross-matching between the three bands, we were left with an average of ∼360 candidates per deg². With ∼4000 deg² in g, r and i, this amounts to ∼1.5 million candidates that form our GOBLIN (Galaxies OBserved as Low-luminosity Identified Nebulae) catalog. For the final classification of these candidates, we finetuned the deep learning model ZOOBOT, which was pretrained based on labels from the Galaxy Zoo project. We created our training dataset by visually inspecting dwarf galaxy candidates from existing literature catalogs within our survey area and assigning probability labels based on averaged expert assessments. This approach captures both consensus and uncertainty among experts. When applied to all detected MTO objects, our method identified 42 965 dwarf galaxy candidates with probability scores of >0.8, of which 23 072 have probabilities exceeding 0.9. The spatial distribution of high-probability candidates reveals a correlation with the locations of massive galaxies (log (M_∗/M_⊙)≥ 10) within 120 Mpc. While some of these objects may have been previously identified in other surveys, we present this extensive catalog of candidates, including their positions, structural parameter estimates, and classification probabilities, as a resource for the community to enable studies of galaxy formation, evolution, and the distribution of dwarf galaxies in different environments.