The second phase of the challenge, focused on characterisation of exoplanets, is on-going (see dedicated tab above) !
DEADLINE EXTENDED TO 30/06/2023 - midnight !
The characterisation phase datasets are hosted on Zenodo
The characterisation phase challenge is hosted on EvalAI
Welcome to The Exoplanet Imaging Data Challenge, which aims at benchmarking and comparing post-processing techniques to detect and characterize exoplanets within ground-based 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 understanding the demographics, the evolution and formation of extrasolar planets. Observing exoplanets using the largest ground-based telescopes is a very challenging task! The main difficulties are (i) the huge difference in brightness between the host star and its potential companions, which are millions of times fainter, and (ii) the small angular separation between them, which are less than a few hundreds milli-arcseconds. Astronomers need the largest optical telescopes to adress this challenge. But, on Earth, these difficulties are amplified by the image degradation caused by the Earth's turbulent atmosphere and the imperfect optics in the instrument. Therefore, ground-based high-contrast imaging (HCI) relies on the use of adaptive optics, which corrects in real time for the atmosphere turbulence by using a deformable mirror, and coronagraphy, which suppresses most of the light coming from the host star.
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 as the exoplanet signal and often brighter by two orders of magnitude. Therefore, the two remaining components of high-contrast imaging are (1) the data acquisition techniques, which focuses on introducing some diversity in the data to later disentangle between the planetary companions and the speckle field and (2) the post-processing techniques that exploit this diversity to carve out further the starlight residuals and recover the planetary signals. Post-processing techniques is what ultimately pushes the performance of the HCI instruments.
Computer science and machine learning fields have a long tradition conducting data challenges and competitions. We aim to integrate these practices to the field of high-contrast imaging and make available repositories of benchmark (curated) datasets to the community. Hopefully in the future, the process of testing new algorithms will be much straightforward and robust, once the community adopts the standard metrics (with their open-source implementations) and the benchmark library resulting from this challenge.
The Exoplanet Imaging Data Challenge is split in different phases conducted every 1-2 year: detection of point sources, characterisation of point sources, detection of extended sources, usage of metadata/telemetry, usage of a reference library, high-resolution spectroscopy data, etc.