J/ApJS/279/26 Morpho-phot. classification of KiDS DR5 sources (Feng+, 2025)
Morpho-photometric classification of KiDS DR5 sources based on neural networks:
a comprehensive star-quasar-galaxy catalog.
Feng H.-C., Li R., Napolitano N.R., Li S.-S., Bai J.M., Dong Y., Li R.,
Liu H.T., Lu K.-X., Pan Z.-W., Radovich M., Shan H.-Y., Wang J.-G.,
Xi W.-Z., Xie L.-H., Yuan Z.-L., Zhang Y.-W.
<Astrophys. J. Suppl. Ser., 279, 26 (2025)>
=2025ApJS..279...26F 2025ApJS..279...26F
ADC_Keywords: Galaxies; QSOs; Infrared sources; Optical; Surveys
Keywords: Neural networks ; Classification ; Surveys ; Catalogs ; A stars ;
Quasars ; Galaxies
Abstract:
We present a novel multimodal neural network (MNN) for classifying
astronomical sources in multiband ground-based observations, from
optical to near-infrared (NIR), to separate sources in stars,
galaxies, and quasars. Our approach combines a convolutional neural
network branch for learning morphological features from r-band images
with an artificial neural network branch for extracting spectral
energy distribution (SED) information. Specifically, we have used
nine-band optical (ugri) and NIR (ZYHJKs) data from the Kilo-Degree
Survey (KiDS) Data Release 5. The two branches of the network are
concatenated and feed into fully connected layers for final
classification. We train the network on a spectroscopically confirmed
sample from the Sloan Digital Sky Survey crossmatched with KiDS. The
trained model achieves 98.76% overall accuracy on an independent
testing data set, with F1-scores exceeding 95% for each class. Raising
the output probability threshold, we obtain higher purity at the cost
of lower completeness. We have also validated the network using
external catalogs crossmatched with KiDS, correctly classifying 99.74%
of a pure star sample selected from Gaia parallaxes and proper
motions, and 99.74% of an external galaxy sample from the Galaxy and
Mass Assembly survey, adjusted for low-redshift contamination. We
apply the trained network to 27,335,836 KiDS DR5 sources with r≤23mag
to generate a new classification catalog. This MNN successfully
leverages both morphological and SED information to enable efficient
and robust classification of stars, quasars, and galaxies in large
photometric surveys.
Description:
KiDS is a ugri-band wide-field imaging survey with the OmegaCAM camera
mounted at the VLT Survey Telescope. In this work, we have developed
an multimodal neural network (MNN) architecture that effectively
combines morphological information from images and multiband
photometry for the classification of stars, quasars, and galaxies in
the KiDS DR5 survey.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table3.dat 73 27335836 KiDS DR5 Galaxy-QSO-Star catalog
(updated version from cosviewer.com)
table4.dat 27 167009 KiDS DR5 training set (see Appendix B)
--------------------------------------------------------------------------------
See also:
VII/233 : 2MASS All-Sky Extended Source Catalog (XSC) (IPAC/UMass, 2003-2006)
II/343 : VIKING catalogue data release 2 (Edge+, 2016)
VII/289 : SDSS quasar catalog, sixteenth data release (DR16Q) (Lyke+, 2020)
II/369 : VEXAS DR2 catalogs (Khramtsov+, 2021)
III/286 : APOGEE-2 DR17 final allStar catalog (Abdurro'uf+, 2022)
I/355 : Gaia DR3 Part 1. Main source (Gaia Collaboration, 2022)
I/357 : Gaia DR3 Part 3. Non-single stars (Gaia Collaboration, 2022)
II/383 : The ESO KiDS-DR5 multi-band source catalog (Wright+ 2024)
J/ApJS/95/1 : Atlas of Quasar Energy Distributions (Elvis+ 1994)
J/ApJ/690/1236 : COSMOS photometric redshift catalog (Ilbert+, 2009)
J/MNRAS/413/971 : Galaxy And Mass Assembly (GAMA) DR1 (Driver+, 2011)
J/MNRAS/483/529 : Star-galaxy multi narrow-band classif. (Cabayol+, 2019)
J/A+A/632/A56 : KiDS-BEXGO catalog (Khramtsov+, 2019)
J/A+A/624/A13 : KiDS DR3 QSO catalog (Nakoneczny+, 2019)
J/A+A/633/A154 : HDBSCAN star, galaxy, QSO classification (Logan+, 2020)
J/MNRAS/496/4276 : Spin parity of spiral galaxies II (Tadaki+, 2020)
J/ApJS/256/34 : SDSS Seyfert 1.9 galaxies with Ha line fit (Chen, 2021)
J/ApJ/909/18 : Reverberation mapping monitoring of NGC3516 (Feng+, 2021)
J/A+A/649/A81 : KiDSDR4 QSOs photometric redshifts cat. (Nakoneczny+, 2021)
J/MNRAS/513/439 : GAMA DR4 spectrum catalog (Driver+, 2022)
J/A+A/661/A3 : eFEDS counterparts to point-like sources (Salvato+, 2022)
J/A+A/660/A34 : X-Shooter Spectral Library (XSL). DR3 (Verro+, 2022)
J/ApJS/264/9 : Southern phot. QSOs from DES DR2 & NIR surveys (Yang+, 2023)
Byte-by-byte Description of file: table3.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 4 A4 --- --- [KiDS]
5- 22 A18 --- ID KiDS DR5 unique identifier for each source
(JHHMMSS.s+DDMMSS.s;
in Simbad)
24- 33 E10.6 deg RAdeg Right ascension in decimal degrees (J2000)
35- 44 E10.6 deg DEdeg [-35.7/4] Declination in decimal degrees (J2000)
46- 55 F10.7 mag rmag [13.4/23] KiDS DR5 r-band magnitude (MAGGAAPr)
57- 62 A6 --- Class Classified as a star, galaxy, or QSO (1)
64- 73 F10.8 --- Pmax [0.33/1] Maximum output probability
--------------------------------------------------------------------------------
Note (1): Classification as follows:
GALAXY = 17,981,882 occurrences
STAR = 6,549,300 occurrences
QSO = 2,804,654 occurrences
--------------------------------------------------------------------------------
Byte-by-byte Description of file: table4.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 10 F10.6 deg RAdeg [128.5/237.3] Right ascension (J2000)
12- 20 F9.6 deg DEdeg [-3.8/4] Declination (J2000)
22- 27 A6 --- Class Classified as a star, galaxy, or QSO (1)
--------------------------------------------------------------------------------
Note (1): Classification as follows:
GALAXY = 116,095 occurrences
STAR = 31,306 occurrences
QSO = 19,608 occurrences
--------------------------------------------------------------------------------
Acknowledgement:
Haicheng Feng [hcfeng at ynao.ac.cn]
History:
From electronic version of the journal
* Table 3 updated by the author (standardized KiDS names and added
r-band magnitudes) downloaded at:
https://cosviewer.com/datasets/kids-dr5-target-classify
(End) Prepared by [AAS], Emmanuelle Perret [CDS] 24-Feb-2026