Spacecraft Thermal Infrared (STIR) Dataset: Annotated IR Imagery for 3DOF Pose Estimation in Proximity Operations

The Spacecraft Thermal Infrared (STIR) Dataset is used primarily to train machine learning-based pose determination methods applied to spacecraft proximity operations. STIR consists of 16,641 thermal infrared pictures of a free-floating spacecraft target platform, captured at Carleton University's Spacecraft Proximity Operations Testbed-an air-bearing laboratory testbed- by an ICI-9320 thermal camera installed on a chaser spacecraft platform. For each of the frames, the relative planar three-degree-of-freedom pose of the target with respect to the chaser is included (x and y positions, and yaw angle). This ground truth data was acquired by a PhaseSpace motion capture system with a submillimeter accuracy.