For the satellite pose estimate machine learning competition, Stanford’s Space Rendezvous Laboratory has generated fully realistic equivalents, using a combination of fully digital images and physical photos taken using satellite models in representative lighting conditions.