ASRU Astral Survey
Research Union

Research

We build open tools that help people look more carefully at astronomical images. Our work is done by volunteers, checked by other volunteers, and published in full.

The “Cosmic Cliffs” of the Carina Nebula, imaged by the James Webb Space Telescope. Image: NASA, ESA, CSA, STScI, public domain

Programs

Plate Archive Project

Classifying stars, emulsion defects and handwritten annotations on digitized historical photographic plates.

Member Image Pipeline

Calibrating, plate-solving and checking the images members take with their own equipment.

Satellite Trails

trailmask, an open model that finds and masks satellite trails in wide-field images.

How an image moves through our work

Each program uses some or all of these steps, in this order.

  1. 01

    Ingest

    Images arrive from public archives or from members, with whatever information comes with them.

  2. 02

    Calibrate

    Instrument effects are removed, using calibration frames where they exist.

  3. 03

    Solve

    Each image is matched to star catalogs so that every pixel has a position on the sky.

  4. 04

    Detect

    Sources, trails and artifacts are found and measured.

  5. 05

    Classify

    Models sort detections into classes and give each a confidence.

  6. 06

    Review

    Members check samples of the results, especially where the model is unsure.

How we work

Open by default

Code is released under the MIT license. Labels, catalogs, model weights and our own text and graphics are released under CC BY 4.0, unless a data source requires otherwise.

People check the machines

No model output is published without a measured comparison against human review. Every technical note reports where the model fails, not only where it works.

Public data, credited properly

We work with publicly released data, including the DASCH plate archive and the Gaia catalog, and acknowledge each source in the way it asks. We are not affiliated with the institutions that publish them.

Tools

Our pipelines are written in Python and use Astropy, ccdproc, SEP, the astrometry.net solver and PyTorch.

Technical notes

All technical notes