Exoplanet Imaging Data Challenge

The second phase of the challenge, focused on characterisation of exoplanets, is on-going (see dedicated tab above) !

DEADLINE EXTENDED TO 31/05/2024 - 23h59 CEST !

The characterisation phase datasets are hosted on Zenodo

The characterisation phase challenge is hosted on EvalAI

To discuss details about the Data Challenge, please join our Discord Server

Welcome to the Exoplanet Imaging Data Challenge, which aims at benchmarking and comparing post-processing techniques applied to high-contrast images. This challenge is organized collaboratively by several members of the astronomical high-contrast imaging community.

Direct imaging of planetary systems is a major step towards studying the demographics of extrasolar planets, as well as understanding the process of formation and evolution of planetary systems. Direct imaging aims at young distant gaseous planets and allows to capture the light emitted by the planet towards analysing its atmosphere properties. However, these objects are very faint and observing exoplanets is a very challenging task. The main difficulties are: (i) the high contrast, i.e. the enormous difference in luminosity between the host star and its planets, which are millions of times less luminous, and (ii) the small angular separation between the star and its planets, which are less than a few hundreds milli-arcseconds.

As a consequence, astronomers need the largest telescopes to adress this challenge, such as ground-based 8-m class telescopes, offering both a good sensitivity and spatial resolution. However, the theoretical spatial resolution of the telescope is hindered by the optical turbulence of the Earth atmosphere. The use of an adaptive optic system (that corrects in real time for the atmosphere turbulence via a deformable mirror) is necessary to reach an angular resolution allowing to probe the inner regions surrounding a star. As a second step, one may use a coronagraph device (an optical component that suppresses most of the light coming from the host star) to reach a contrast allowing to detect the planet signal that is million times fainter than the starlight.

Even using these cutting-edge technologies, the resulting images suffer from bright starlight residuals hiding the presence of planetary companions. Among them, the quasi-statics speckles are of the same size and shape as the exoplanet signal and often brighter by up to two orders of magnitude. A necessary last step is to apply tailored image processing techniques to the dataset, which further carve out the residual starlight and unveil the faint circumstellar signals (disks and/or planets). The image below highlights the "four pillars" of high-contrast imaging needed to detect Jupiter-sized exoplanets located at large distances (a few tenth of astronomical units) from their host star.

Computer science and artificial intelligence fields have a long tradition conducting data challenges and methods competitions. We aim to integrate these practices to the field of high-contrast imaging in astronomy. The Exoplanet Imaging Data Challenge (EIDC) intends to make the process of testing new algorithms more straightforward and robust, via the use of standard metrics and benchmark (curated) data set offered to the community with this challenge.

In the future, it is planned to regularly launch several phases of the Exoplanet Imaging Data Challenge, and make it evolve to include more features. Next steps will notably include the detection of circumstellar disks, the use of high resolving power spectroscopic data, the use of large data base and multi-epoch, and simulated images for the Extremely Large Telescope suites of instrument.

The EIDC team plans to publish its work and various results every two years, under the form of an SPIE Astronomical Telescope + Instrumentation proceeding that will be made available on arXiv (see Results tabs).