AKM Dataset: Textureless Space Target Dataset

The target in this dataset is the Apogee Kick Motor (AKM) (similar to the STAR 24 motor), a radially symmetric and textureless space object. The motor dimensions are a diameter of 24.5 inches (0.62 m) and a length of 42.0 inches (1.0668 m). The dataset contains 10000 images generated with a blender under lighting conditions similar to the orbital environment with an image size of 1024 x 1024 pixels rendered using a pin-hole camera model having a field of view of 55 degrees. The target positions are obtained from a uniform random distribution between 1m and 10 m, and orientations are obtained from a uniform random distribution.

The groundtruth.csv contains the pose labels.
filename - filename of the image.
Tx, Ty, Tz - represents the location of the target in the Camera coordinate frame
Qx, Qy, Qz, Qw - quaternion representation of target orientation in camera coordinate frame
K.txt - contains the Camera intrinsic matrix.

@inproceedings{rathinam2022pose, title={Pose Estimation of a Known Texture-Less Space Target using Convolutional Neural Networks}, author={Rathinam, Arunkumar and Gaudilliere, Vincent and Pauly, Leo and Aouada, Djamila}, booktitle={73rd International Astronautical Congress, Paris 18-22 September 2022}, year={2022} }