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 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 $\sim 30$ 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 LSST and WFIRST. Our code, Astro R-CNN, is publicly available at https://github.com/burke86/astro_rcnn.

13 pages, 12 figures, accepted for publication in MNRAS

References in corpus (47)

• Adam: A Method for Stochastic Optimization
• Astropy: A Community Python Package for Astronomy
• The Astropy Project: Building an inclusive, open-science project and status of the v2.0 core package
• On the variation of the Initial Mass Function
• Nine-Year Wilkinson Microwave Anisotropy Probe (WMAP) Observations: Cosmological Parameter Results
• Cosmic Star Formation History
• LSST: from Science Drivers to Reference Design and Anticipated Data Products
• The clustering of galaxies in the SDSS-III Baryon Oscillation Spectroscopic Survey: Baryon Acoustic Oscillations in the Data Release 10 and 11 galaxy samples
• Overview of the DESI Legacy Imaging Surveys
• The ages and metallicities of galaxies in the local universe
• The Dark Energy Camera
• Wide-Field InfrarRed Survey Telescope-Astrophysics Focused Telescope Assets WFIRST-AFTA 2015 Report
• Rotation-invariant convolutional neural networks for galaxy morphology prediction
• First Data Release of the Hyper Suprime-Cam Subaru Strategic Program
• TernausNet: U-Net with VGG11 Encoder Pre-Trained on ImageNet for Image Segmentation
• Cosmology from cosmic shear power spectra with Subaru Hyper Suprime-Cam first-year data
• New Grids of ATLAS9 Model Atmospheres
• The Dark Energy Survey Data Release 1
• Preparing Red-Green-Blue (RGB) Images from CCD Data
• Path Aggregation Network for Instance Segmentation
• S4N: A Spectroscopic Survey of Stars in the Solar Neighborhood
• Star-galaxy Classification Using Deep Convolutional Neural Networks
• Photometric redshifts from SDSS images using a Convolutional Neural Network
• Ages of galaxy bulges and disks from optical and near-infrared colors
• The atmospheric parameters and spectral interpolator for the stars of MILES
• Photometric redshift estimation via deep learning
• The Effective Number Density of Galaxies for Weak Lensing Measurements in the LSST Project
• PopStar I: Evolutionary synthesis models description
• Gaia DR2 unravels incompleteness of nearby cluster population: New open clusters in the direction of Perseus
• Optimising Automatic Morphological Classification of Galaxies with Machine Learning and Deep Learning using Dark Energy Survey Imaging
• Measurement of the Mass Profile of Abell 1689
• Characterization and Photometric Performance of the Hyper Suprime-Cam Software Pipeline
• SCARLET: Source separation in multi-band images by Constrained Matrix Factorization
• Decision Tree Classifiers for Star/Galaxy Separation
• Gravitational Lens Magnification and the Mass of Abell 1689
• Faster Training of Mask R-CNN by Focusing on Instance Boundaries
• Simulation of Astronomical Images from Optical Survey Telescopes using a Comprehensive Photon Monte Carlo Approach
• Wide Field Imaging. I. Applications of Neural Networks to object detection and star/galaxy classification
• Star/galaxy separation at faint magnitudes: Application to a simulated Dark Energy Survey
• Photometry of high-redshift blended galaxies using deep learning
• The Ellipticity Distribution of Ambiguously Blended Objects
• Deblending galaxy superpositions with branched generative adversarial networks
• GROWTH on S190426c. II. Real-Time Search for a Counterpart to the Probable Neutron Star-Black Hole Merger using an Automated Difference Imaging Pipeline for DECam
• A Hybrid Ensemble Learning Approach to Star-Galaxy Classification
• Star-galaxy classification in the Dark Energy Survey Y1 dataset
• Crowded Cluster Cores: Algorithms for Deblending in Dark Energy Survey Images
• Mask Scoring R-CNN

Cited by in corpus (32)

• The DAWES review 10: The impact of deep learning for the analysis of galaxy surveys
• Detection and Classification of Astronomical Targets with Deep Neural Networks in Wide Field Small Aperture Telescopes
• Automated galaxy-galaxy strong lens modelling: no lens left behind
• Discovery and analysis of low-surface-brightness galaxies in the environment of NGC 1052
• Detection, Instance Segmentation, and Classification for Astronomical Surveys with Deep Learning (DeepDISC): Detectron2 Implementation and Demonstration with Hyper Suprime-Cam Data
• The search for galaxy cluster members with deep learning of panchromatic HST imaging and extensive spectroscopy
• SILVERRUSH X: Machine Learning-Aided Selection of $9318$ LAEs at $z=2.2$, $3.3$, $4.9$, $5.7$, $6.6$, and $7.0$ from the HSC SSP and CHORUS Survey Data
• Radio astronomical images object detection and segmentation: A benchmark on deep learning methods
• YOLO-CIANNA: Galaxy detection with deep learning in radio data. I. A new YOLO-inspired source detection method applied to the SKAO SDC1
• Point Spread Function Estimation for Wide Field Small Aperture Telescopes with Deep Neural Networks and Calibration Data
• Radio Sources Segmentation and Classification with Deep Learning
• The miniJPAS survey quasar selection II: Machine learning classification with photometric measurements and uncertainties
• Gaussian Process Classification for Galaxy Blend Identification in LSST
• The miniJPAS survey quasar selection III: Classification with artificial neural networks and hybridisation
• A Comparison of Deep Learning Architectures for Optical Galaxy Morphology Classification
• The Evryscope Fast Transient Engine: Real-Time Detection for Rapidly Evolving Transients
• The Classification of Optical Galaxy Morphology Using Unsupervised Learning Techniques
• Score-matching neural networks for improved multi-band source separation
• Accurately Estimating Redshifts from CSST Slitless Spectroscopic Survey using Deep Learning
• AGNet: Weighing Black Holes with Deep Learning
• Eliminating artefacts in Polarimetric Images using Deep Learning
• How to set up your first machine learning project in astronomy
• GalaxyGenius: Mock galaxy image generator for various telescopes from hydrodynamical simulations
• Automated galaxy sizes in Euclid images using the Segment Anything Model
• Improved source classification and performance analysis using Gaia DR3
• AGNet: Weighing Black Holes with Machine Learning
• 200,000+ Deep Learning-inferred Periods of Stellar Variability from the All-Sky Automated Survey for Supernovae
• Nonparametric signal separation in very-high-energy gamma ray observations with probabilistic neural networks
• ComPACT: ACT+Planck galaxy cluster catalogue
• CSST Slitless Spectra: Target Detection and Classification with YOLO
• Smart obervation method with wide field small aperture telescopes for real time transient detection
• Predicting large scale cosmological structure evolution with generative adversarial network-based autoencoders