skip to main content

Title: Deblending and classifying astronomical sources with Mask R-CNN deep learning
ABSTRACT

We apply a new deep learning technique to detect, classify, and deblend sources in multiband astronomical images. We train and evaluate the performance of an artificial neural network built on the Mask Region-based Convolutional Neural Network image processing framework, a general code for efficient object detection, classification, and instance segmentation. After evaluating the performance of our network against simulated ground truth images for star and galaxy classes, we find a precision of 92 per cent at 80 per cent recall for stars and a precision of 98 per cent at 80 per cent recall for galaxies in a typical field with ∼30 galaxies arcmin−2. We investigate the deblending capability of our code, and find that clean deblends are handled robustly during object masking, even for significantly blended sources. This technique, or extensions using similar network architectures, may be applied to current and future deep imaging surveys such as Large Synoptic Survey Telescope and Wide-Field Infrared Survey Telescope. Our code, astro r-cnn, is publicly available at https://github.com/burke86/astro_rcnn.

Authors:
 ;  ;  ;  ;  ;  ;  
Publication Date:
NSF-PAR ID:
10123129
Journal Name:
Monthly Notices of the Royal Astronomical Society
Volume:
490
Issue:
3
Page Range or eLocation-ID:
p. 3952-3965
ISSN:
0035-8711
Publisher:
Oxford University Press
Sponsoring Org:
National Science Foundation
More Like this
  1. e apply a new deep learning technique to detect, classify, and deblend sources in multi-band astronomical images. We train and evaluate the performance of an artificial neural network built on the Mask R-CNN image processing framework, a general code for efficient object detection, classification, and instance segmentation. After evaluating the performance of our network against simulated ground truth images for star and galaxy classes, we find a precision of 92% at 80% recall for stars and a precision of 98% at 80% recall for galaxies in a typical field with ∼30 galaxies/arcmin2. We investigate the deblending capability of our code, and find that clean deblends are handled robustly during object masking, even for significantly blended sources. This technique, or extensions using similar network architectures, may be applied to current and future deep imaging surveys such as LSST and WFIRST. Our code, Astro R-CNN, is publicly available at https://github.com/burke86/astro_rcnn
  2. ABSTRACT We present the results of a proof-of-concept experiment that demonstrates that deep learning can successfully be used for production-scale classification of compact star clusters detected in Hubble Space Telescope(HST) ultraviolet-optical imaging of nearby spiral galaxies ($D\lesssim 20\, \textrm{Mpc}$) in the Physics at High Angular Resolution in Nearby GalaxieS (PHANGS)–HST survey. Given the relatively small nature of existing, human-labelled star cluster samples, we transfer the knowledge of state-of-the-art neural network models for real-object recognition to classify star clusters candidates into four morphological classes. We perform a series of experiments to determine the dependence of classification performance on neural network architecture (ResNet18 and VGG19-BN), training data sets curated by either a single expert or three astronomers, and the size of the images used for training. We find that the overall classification accuracies are not significantly affected by these choices. The networks are used to classify star cluster candidates in the PHANGS–HST galaxy NGC 1559, which was not included in the training samples. The resulting prediction accuracies are 70 per cent, 40 per cent, 40–50 per cent, and 50–70 per cent for class 1, 2, 3 star clusters, and class 4 non-clusters, respectively. This performance is competitive with consistency achieved in previously published human and automated quantitative classification of starmore »cluster candidate samples (70–80 per cent, 40–50 per cent, 40–50 per cent, and 60–70 per cent). The methods introduced herein lay the foundations to automate classification for star clusters at scale, and exhibit the need to prepare a standardized data set of human-labelled star cluster classifications, agreed upon by a full range of experts in the field, to further improve the performance of the networks introduced in this study.« less
  3. ABSTRACT Strongly lensed quadruply imaged quasars (quads) are extraordinary objects. They are very rare in the sky and yet they provide unique information about a wide range of topics, including the expansion history and the composition of the Universe, the distribution of stars and dark matter in galaxies, the host galaxies of quasars, and the stellar initial mass function. Finding them in astronomical images is a classic ‘needle in a haystack’ problem, as they are outnumbered by other (contaminant) sources by many orders of magnitude. To solve this problem, we develop state-of-the-art deep learning methods and train them on realistic simulated quads based on real images of galaxies taken from the Dark Energy Survey, with realistic source and deflector models, including the chromatic effects of microlensing. The performance of the best methods on a mixture of simulated and real objects is excellent, yielding area under the receiver operating curve in the range of 0.86–0.89. Recall is close to 100 per cent down to total magnitude i ∼ 21 indicating high completeness, while precision declines from 85 per cent to 70 per cent in the range i ∼ 17–21. The methods are extremely fast: training on 2 million samples takes 20 h on a GPU machine, and 108more »multiband cut-outs can be evaluated per GPU-hour. The speed and performance of the method pave the way to apply it to large samples of astronomical sources, bypassing the need for photometric pre-selection that is likely to be a major cause of incompleteness in current samples of known quads.« less
  4. Abstract

    Rejecting cosmic rays (CRs) is essential for the scientific interpretation of CCD-captured data, but detecting CRs in single-exposure images has remained challenging. Conventional CR detectors require experimental parameter tuning for different instruments, and recent deep-learning methods only produce instrument-specific models that suffer from performance loss on telescopes not included in the training data. We present Cosmic-CoNN, a generic CR detector deployed for 24 telescopes at the Las Cumbres Observatory, which has been made possible by the three contributions in this work: (1) We build a large and diverse ground-based CR data set leveraging thousands of images from a global telescope network. (2) We propose a novel loss function and a neural network optimized for telescope imaging data to train generic CR-detection models. At 95% recall, our model achieves a precision of 93.70% on Las Cumbres imaging data and maintains a consistent performance on new ground-based instruments never used for training. Specifically, the Cosmic-CoNN model trained on the Las Cumbres CR data set maintains high precisions of 92.03% and 96.69% on Gemini GMOS-N/S 1 × 1 and 2 × 2 binning images, respectively. (3) We build a suite of tools including an interactive CR mask visualization and editing interface, consolemore »commands, and Python APIs to make automatic, robust CR detection widely accessible by the community of astronomers. Our data set, open-source code base, and trained models are available athttps://github.com/cy-xu/cosmic-conn.

    « less
  5. ABSTRACT

    Tidal features in the outskirts of galaxies yield unique information about their past interactions and are a key prediction of the hierarchical structure formation paradigm. The Vera C. Rubin Observatory is poised to deliver deep observations for potentially millions of objects with visible tidal features, but the inference of galaxy interaction histories from such features is not straightforward. Utilizing automated techniques and human visual classification in conjunction with realistic mock images produced using the NewHorizon cosmological simulation, we investigate the nature, frequency, and visibility of tidal features and debris across a range of environments and stellar masses. In our simulated sample, around 80 per cent of the flux in the tidal features around Milky Way or greater mass galaxies is detected at the 10-yr depth of the Legacy Survey of Space and Time (30–31 mag arcsec−2), falling to 60 per cent assuming a shallower final depth of 29.5 mag arcsec−2. The fraction of total flux found in tidal features increases towards higher masses, rising to 10 per cent for the most massive objects in our sample (M⋆ ∼ 1011.5 M⊙). When observed at sufficient depth, such objects frequently exhibit many distinct tidal features with complex shapes. The interpretation and characterization of such features varies significantly withmore »image depth and object orientation, introducing significant biases in their classification. Assuming the data reduction pipeline is properly optimized, we expect the Rubin Observatory to be capable of recovering much of the flux found in the outskirts of Milky Way mass galaxies, even at intermediate redshifts (z < 0.2).

    « less