Detecting solar system objects with convolutional neural networks

Abstract

In the preparation for ESA's Euclid mission and the large amount of data it will produce, we train deep convolutional neural networks on Euclid simulations classify solar system objects from other astronomical sources. Using transfer learning we are able to achieve a good performance despite our tiny dataset with as few as 7512 images. Our best model correctly identifies objects with a top accuracy of 94% and improves to 96% when Euclid's dither information is included. The neural network misses ~50% of the slowest moving asteroids (v < 10 arcsec/h) but is otherwise able to correctly classify asteroids even down to 26 mag. We show that the same model also performs well at classifying stars, galaxies and cosmic rays, and could potentially be applied to distinguish all types of objects in the Euclid data and other large optical surveys.

11 pages, 10 figures, accepted for publication in MNRAS

References in corpus (24)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• Gaia Data Release 2. Summary of the contents and survey properties
• Neural Architecture Search with Reinforcement Learning
• KiDS-450: Cosmological parameter constraints from tomographic weak gravitational lensing
• LSST Science Book, Version 2.0
• Euclid Definition Study Report
• Rotation-invariant convolutional neural networks for galaxy morphology prediction
• Solar System evolution from compositional mapping of the asteroid belt
• The taxonomic distribution of asteroids from multi-filter all-sky photometric surveys
• Xception: Deep Learning with Depthwise Separable Convolutions
• Origin of water in the inner Solar System: Planetesimals scattered inward during Jupiter and Saturn's rapid gas accretion
• Photometric redshifts from SDSS images using a Convolutional Neural Network
• Asteroid family ages
• Using transfer learning to detect galaxy mergers
• The Euclid mission design
• Transfer learning for galaxy morphology from one survey to another
• Dynamical Analysis of Nearby ClustErs. Automated astrometry from the ground: precision proper motions over wide field
• Deep Convolutional Neural Networks as strong gravitational lens detectors
• VIS: the visible imager for Euclid
• Solar System Science with ESA Euclid
• Hyperparameters Optimization in Deep Convolutional Neural Network / Bayesian Approach with Gaussian Process Prior
• Why is Interstellar Object 1I/2017 U1 (`Oumuamua) Rocky, Tumbling and Very Prolate?
• Mining the Kilo-Degree Survey for solar system objects

Cited by in corpus (18)

• DeepStreaks: identifying fast-moving objects in the Zwicky Transient Facility data with deep learning
• CASI: A Convolutional Neural Network Approach for Shell Identification
• Hubble Asteroid Hunter: I. Identifying asteroid trails in Hubble Space Telescope images
• Towards Efficient Detection of Small Near-Earth Asteroids Using the Zwicky Transient Facility (ZTF)
• Discovering Faint and High Apparent Motion Rate Near-Earth Asteroids Using A Deep Learning Program
• Tracing Halpha Fibrils through Bayesian Deep Learning
• Euclid: Identification of asteroid streaks in simulated images using deep learning
• Identifying and Tracking Solar Magnetic Flux Elements with Deep Learning
• PGNets: Planet mass prediction using convolutional neural networks for radio continuum observations of protoplanetary disks
• Neural network for determining an asteroid mineral composition from reflectance spectra
• Towards Asteroid Detection in Microlensing Surveys with Deep Learning
• Euclid: Identification of asteroid streaks in simulated images using StreakDet software
• Deep Transfer Learning for Classification of Variable Sources
• The ssos Pipeline: Identification of Solar System Objects in Astronomical Images
• Multiwavelength classification of X-ray selected galaxy cluster candidates using convolutional neural networks
• Deep learning to improve the discovery of near-Earth asteroids in the Zwicky Transient Facility
• Exploring the Solar System with the NOIRLab Source Catalog I: Detecting Objects with CANFind
• StreakMind: AI detection and analysis of satellite streaks in astronomical images with automated database integration