The data for this competition is released under the CC BY-NC-SA 3.0 license, and can be downloaded as a compressed zip file (4.6 GB).
The zip file containing the dataset is organized as follows:
train/: a folder containing \( 12000 \) synthetic images for training. Images are 8 bit monochrome in jpeg format, with a resolution of \(1920 \times 1200\) pixels.
test/: \( 2998 \) similar synthetic images for evaluating submissions
real_test/: \( 300 \) real images of the Tango satellite mock-up, same format and resolution as the synthetic images.
real/: \( 5\) example real images, with pose labels
train.json: filenames and corresponding pose labels for the 12000 training images
test.json: list of filenames of the test images
real_test.json: list of filenames of the real test images
real.json: filenames and corresponding pose labels for five example real image
LICENSE.MD: dataset license file
We provide teams with utility scripts that help working with the dataset and support exporting submissions. This
speed-utils repository can be cloned here: https://gitlab.com/EuropeanSpaceAgency/speed-utils