Published April 22, 2022 | Version Version 1.0
Dataset Open

Exoplanet imaging data challenge, phase 2

  • 1. Laboratoire d'Astrophysique de Marseille (LAM, France)
  • 2. Université de Liège (ULiège, Belgium)
  • 3. Max Planck Institute for Astronomy (MPIA, Germany)
  • 1. Université de Liège (ULiège, Belgium)
  • 2. ETH Zurich (Switzerland)
  • 3. Observatoire de Paris (LESIA, France)
  • 4. Max Planck Institute for Astronomy (MPIA, Germany)
  • 5. CalTech (USA)
  • 6. Institut de Planétologie et d'Astrophysique de Grenoble (IPAG, France)


Datasets for the second phase of the Exoplanet Imaging Data Challenge ( 

The second phase of the Exoplanet Imaging Data Challenge is focused on the characterisation of exoplanet signals in high-contrast imaging data. The participants must perform two tasks: provide (1) astrometry of the detected signals, and (2) spectrophotometry of the detected signals.

For this phase, we therefore provide with 8 high-contrast data sets, taken with two integral field spectrographs: SPHERE-IFS installed at the Very Large Telescope (VLT, Chile) and GPI installed at the Gemini-South telescope (Chile). Each data set consists of the following files (in .fits format):

(a) image_cube_instxx.fits: the coronagraphic multispectral image cube, acquired in pupil-stabilized mode;
(b) parallactic_angles_instxx.fits: the corresponding parallactic angle and airmass variation during the observation sequence;
(c) wavelength_vect_instxx.fits: the corresponding wavelength vector for each spectral channel;
(d) psf_cube_instxx.fits: the non-coronagraphic (point spread function) multispectral image of the target star;
(e) first_guess_astrometry_instxx.fits: a first guess position w.r.t the star of the (2 or 3) injected planetary signals.

Each data set has 2 to 3 injected planetary signals in various locations.
Each data set has very different observing conditions, from very good to very bad.

Additional information and ressources can be found on the website and dedicated Github repository.

The results of the data challenge must be submitted directly by participants on the EvalAI platform.


On 22nd of July 2022: we modified the coronagraphic image cube *and* the psf image cube of the 'sphere2' data set because the low wind effect was not treated correctly.


Files (2.8 GB)

Name Size Download all
559.8 MB Preview Download
1.9 GB Preview Download
362.2 MB Download
5.8 kB Download
135.4 kB Download
5.8 kB Download