Data

The data for this competition can be downloaded as a compressed zip file.

Download from Kelvins

Contents

The following files are contained in the zip:

• data/input.csv: This is the Input file. It contains the scenes which have to be identified.
• data/hip_main.dat: This is the Hipparchos catalog.

Other files are also included in the zip and, while not strictly necessary, their purpose is explained on the submission format page.

Input File

The input file is a headerless CSV file with varying column count. Each row/line in the file represents a scene of stars as seen by the camera. The scenes are stored as an even number of comma separated numbers. These are the coordinates (and magnitude) of the stars in the image, as described on the description page. For \(N\) stars in the scene the coordinates are stored as \(x_1\),\(y_1\),\(m_1\),\(x_2\),\(y_2\),\(m_2\),...,\(x_N\),\(y_N\),\(m_N\), for example (values rounded):

397.0,331.7,3.9,490.4,1249.7,5.3,1883.3,637.2,2.7,170.8,13.4,4.4,1902.3,522.3,5.1,1551.5,1013.0,4.9

Hipparchos catalog

The catalog (http://www.cosmos.esa.int/web/hipparcos) is stored as a fixed width text file with the character | as separator. The important values for this competition can be found in columns 2, 6, 9, and 10, which are the HIP number \(h\) between 1 and 120416 (bytes 9-14), the visual magnitude \(m\) (bytes 42-46), the Right Ascension \(\omega\) (byes 52-63) and the Declination \(\psi\) (bytes 65-76) in degrees.

A more detailed description can be found in the VizieR archives for example. Also, the source code available on the scripts page contains a function to load the catalog with Python.

The scenes

The input file contains ten thousand scenes. The parameters used to create such scenes (simulating the star tracker camera) are:

• resolution: 1920x1440 pixels
• normalized principal point \(p_x\), \(p_y\): \(0.5\)
• \(fov_x = 10\) degrees
• \(\frac{s_x}{s_y} = 1\)
• no distortion (rectilinear/perspective camera model)
• centroiding noise typically below 1 pixel
• measurement error on the magnitude measurement (error is increasing at higher magnitudes)
• up to 80 artifacts

See the extra info in the challenge page for a first explanation of these parameters. You can also play around with the example scripts to see the effect of these parameters on the scene created.

Note that some scenes are impossible to be correctly identified.