Detection, instance segmentation, and classification for astronomical surveys with deep learning (deepdisc): detectron2 implementation and demonstration with Hyper Suprime-Cam data

Authors:
Publisher website
Google Scholar
Abstract
The next generation of wide-field deep astronomical surveys will deliver unprecedented amounts of images through the 2020s and beyond. As both the sensitivity and depth of observations increase, more blended sources will be detected. This reality can lead to measurement biases that contaminate key astronomical inferences. We implement new deep learning models available through Facebook AI Research’s Detectron2 repository to perform the simultaneous tasks of object identification, deblending, and classification on large multi-band coadds from the Hyper Suprime-Cam (HSC). We use existing detection/deblending codes and classification methods to train a suite of deep neural networks, including state-of-the-art transformers. Once trained, we find that transformers outperform traditional convolutional neural networks and are more robust to different contrast scalings. Transformers are able to detect and deblend objects closely matching the ground truth, achieving a median bounding box Intersection over Union of 0.99. Using high quality class labels from the Hubble Space Telescope, we find that when classifying objects as either stars or galaxies, the best-performing networks can classify galaxies with near 100% completeness and purity across the whole test sample and classify stars above 60% completeness and 80% purity out to HSC i-band magnitudes of 25 mag. This framework can be extended to other upcoming deep surveys such as the Legacy Survey of Space and Time and those with the Roman Space Telescope to enable fast source detection and measurement. Our code, DeepDISC, is publicly available at https://github.com/grantmerz/deepdisc.
Funding Information
  • NCSA
  • National Science Foundation (1725729, AST-1238877)
  • National Astronomical Observatory of Japan
  • University of Tokyo
  • High Energy Accelerator Research Organization
  • Princeton University
  • Ministry of Education, Culture, Sports, Science and Technology
  • Japan Society for the Promotion of Science
  • Johns Hopkins University
  • University of Edinburgh
  • Space Telescope Science Institute
  • National Aeronautics and Space Administration (NNX08AR22G)
  • Science Mission Directorate
  • California Institute of Technology

This publication has 52 references indexed in Scilit: