J/ApJS/278/16 Star clusters in M31 from PAndAS and deep learning (Zhang+, 2025)
Identification of star clusters in M31 from PAndAS images based on deep
learning.
Zhang B., Chen B., Yuan H., Chen P., Wang S., Zhang L., Ren Yi, Guo H.
<Astrophys. J. Suppl. Ser., 278, 16 (2025)>
=2025ApJS..278...16Z 2025ApJS..278...16Z
ADC_Keywords: Galaxies, nearby; Clusters, open; Photometry, RI
Keywords: Andromeda galaxy ; Young star clusters ;
Convolutional neural networks ; Star clusters
Abstract:
The identification of star clusters holds significant importance in
studying galaxy formation and evolution history. However, the task of
swiftly and accurately identifying star clusters from vast amounts of
photometric images presents an immense challenge. To address these
difficulties, we employ deep learning models for image classification
to identify young disk star clusters in M31 from the Pan-Andromeda
Archaeological Survey (PAndAS) images. For training, validation, and
testing, we utilize the Panchromatic Hubble Andromeda Treasury survey
catalogs. We evaluate the performance of various deep learning models,
using different classification thresholds and limiting magnitudes. Our
findings indicate that the ResNet-50 model exhibits the highest
overall accuracy. Moreover, using brighter limiting magnitudes and
increasing the classification thresholds can effectively enhance the
accuracy and precision of cluster identification. Through our
experiments, we found that the model achieves optimal performance when
the limiting magnitude is set to brighter than 21 mag. Based on this,
we constructed a training data set with magnitudes less than 21 mag
and trained a second ResNet-50 model. This model achieved a purity of
89.30%, a recall of 73.55%, and an F1 score of 80.66% when the
classification threshold was set to 0.669. Applying the second model
to all sources in the PAndAS fields within a projected radius of 30
kpc from the center of M31, we identified 2228 new unique star cluster
candidates. We conducted visual inspections to validate the results
produced by our automated methods, and we ultimately obtained
1057 star cluster candidates, of which 745 are newly identified.
Description:
In this study, we utilize the Pan-Andromeda Archaeological Survey
(PAndAS) g- and i-band images obtained from the CFHT to train our deep
learning models and identify new star cluster candidates in M31. The
survey was conducted between 2003 and 2010, using the MegaCam
wide-field camera.
The camera has an effective field of view of 0.96x0.94deg2 and a pixel
scale of 0.187"/pixel. The typical seeing values for the g and i bands
in the PAndAS survey are 0.67" and 0.60", respectively. In this work,
we use the processed stacked images that were processed by the
Cambridge Astronomical Survey Unit from the PAndAS VOspace.
To train our deep learning models, we require a substantial number of
positive and negative sample images. As positive samples, we utilize
the confirmed star cluster catalog derived from the PHAT survey
project conducted by Johnson+ (2015, J/ApJ/802/127). This catalog was
constructed through the classification of 20,000 images by tens of
thousands of volunteers, resulting in 1.82 million classifications and
the identification of 2753 star clusters.
Objects:
----------------------------------------------------------
RA (ICRS) DE Designation(s)
----------------------------------------------------------
00 42 44.33 +41 16 07.5 M31 = NAME Andromeda Galaxy
----------------------------------------------------------
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table7.dat 128 1057 Catalog of the newly identified 1057 star cluster
candidates in M31
--------------------------------------------------------------------------------
See also:
J/AJ/137/94 : Star clusters in M31 (Caldwell+, 2009)
J/PASP/122/745 : HST WFPC2 star clusters in M31. III. (Hodge+, 2010)
J/AJ/141/61 : Star clusters in M31. II. (Caldwell+, 2011)
J/ApJ/752/95 : PHAT stellar cluster survey. I. Year 1 (Johnson+, 2012)
J/other/RAA/15.1392 : LAMOST globular clusters in M 31 and M 33 (Chen+, 2015)
J/ApJ/802/127 : PHAT stellar cluster survey. II. AP catalog (Johnson+, 2015)
J/ApJ/868/55 : Structure of M31 halo. II. PAndAS (McConnachie+, 2018)
J/A+A/658/A51 : New M31 star cluster candidates (Wang+, 2022)
J/ApJ/954/206 : GC search in M31 with Gaia, PS1, LAMOST, PAndAS (Wang+, 2023)
http://www.cfht.hawaii.edu/Science/PandAS/ : PandAS data access
Byte-by-byte Description of file: table7.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 4 I4 --- Seq [1/1057] Candidate sequential identifier
6- 6 I1 h RAh [0] Hour of Right Ascension (J2000)
8- 9 I2 min RAm [31/50] Minute of Right Ascension (J2000)
11- 16 F6.3 s RAs Second of Right Ascension (J2000)
18- 18 A1 --- DE- Sign of the Declination (J2000)
19- 20 I2 deg DEd [38/43] Degree of Declination (J2000)
22- 23 I2 arcmin DEm Arcminute of Declination (J2000)
25- 29 F5.2 arcsec DEs Arcsecond of Declination (J2000)
31- 36 F6.3 mag gmag [13.5/21] PAndAS CFHT/MegaCam g-band AB
magnitude
38- 42 F5.3 mag e_gmag [0.001/0.03] Uncertainty in gmag
44- 49 F6.3 mag imag [12.74/35.02] PAndAS CFHT/MegaCam i-band AB
magnitude
51- 55 F5.3 mag e_imag [0.001/9.4] Uncertainty in imag
57- 59 I3 --- PAndAS [207/285] PAndAS field ID
61- 65 F5.2 kpc Rproj [1/83] Projected distance to center of M31
67- 72 F6.4 --- Conf [0/1] Classification Confidence (highest
probability = most likely class; Section 3.1)
74- 74 I1 --- Score [3/4] Visual inspection score (1)
76- 92 A17 --- C09 Name in Caldwell et al. 2009, J/AJ/137/94
94-102 A9 --- RBC Name in Galleti et al. 2004, J/A+A/416/917 (2)
104-113 A10 --- C11 Name in Caldwell et al. 2011, J/AJ/141/61
115-116 I2 --- W22 ? Name in Wang et al. 2022, J/A+A/658/A51
118 I1 --- W23 ? Name in Wang et al. 2023, J/ApJ/954/206
120-128 A9 --- H14 Name in Huxor et al. 2014MNRAS.442.2165H 2014MNRAS.442.2165H
--------------------------------------------------------------------------------
Note (1): Visual inspection score system corresponds to the sum of
the following values for two manual inspectors: 2 points for
confirmed star clusters, 1 point for potential star clusters and
0 points for non-clusters. Only scores of 3 or 4 are included.
Note (2): RBC = Galleti et al. 2004, J/A+A/416/917; superseded by V/143;
http://www.bo.astro.it/M31/
--------------------------------------------------------------------------------
History:
From electronic version of the journal
(End) Prepared by [AAS], Emmanuelle Perret [CDS] 12-Feb-2026