Morpheus: A Deep Learning Framework For Pixel-Level Analysis of Astronomical Image Data

Ryan Hausen and Brant Robertson
University Of California, Santa Cruz
ArXiv

We present Morpheus, a new model for generating pixel level morphological classifications of astronomical sources. Morpheus leverages advances in deep learning to perform source detection, source segmentation, and morphological classification pixel-by-pixel via a semantic segmentation algorithm adopted from the field of computer vision. By utilizing morphological information about the flux of real astronomical sources during object detection, Morpheus shows resiliency to false positive identifications of sources. We evaluate Morpheus by performing source detection, source segmentation, morphological classification on the Hubble Space Telescope data in the GOODS South field, and demonstrate a high completeness in recovering known 3D-HST sources with H < 26 AB. We release the code publicly, provide online demonstrations, and present an interactive visualization of the Morpheus results in GOODS South.

Code Release

Morpheus is available via Github, pip, and Docker. Try it out in a simple tutorial!

Data Release

The following data products are released with the paper:

File Description
morpheus_GOODS-S_spheroid.v1.0.fits The pixel-level classifications for the spheroid class
morpheus_GOOD-S_disk.v1.0.fits The pixel-level classifications for the disk class
morpheus_GOOD-S_irregular.v1.0.fits The pixel-level classifications for the irregular class
morpheus_GOOD-S_ps_compact.v1.0.fits The pixel-level classifications for the point source/compact class
morpheus_GOOD-S_background.v1.0.fits The pixel-level classifications for the background class
morpheus_GOOD-S_mask.v1.0.fits Binary pixel map indicating classified pixels in the survey
morpheus GOODS-S_3dhst_segmap.v1.0.fits Segmentation map generated using Morpheus and the 3DHST source locations
morpheus_GOODS-S_segmap.v1.0.fits Segmentation map generated using Morpheus
morpheus_GOODS-S_3dhst_catalog.v1.0.csv Morphological classifications for objects in the 3DHST Survey (CSV)
morpheus_GOODS-S_3dhst_catalog.v1.0.txt Morphological classifications for objects in the 3DHST Survey (Machine Readable Table)
morpheus_GOODS_morph_color.v1.0.png.tar.gz Color image composite of morphological classifications
morpheus_GOODS-S_all.v1.0.tar.gz All of the above files compressed together

Examples

zoom

Interactive Map

The data release can be explored through two interactive maps. The first, a light-weight version contains an RGB composite image based on the H, Z, and V bands respectively and the Morpheus per-pixel classification. The second contains additional information, including:

  • Segmentation maps
  • 3D-HST source locations with a search tool
  • Map layers for each morphological class
  • Map layers for each input band of light