Datasets:
metadata
license: cc-by-4.0
pretty_name: 'ASTRA: Autonomous Spacecraft Tracking and Rendezvous Archive'
task_categories:
- keypoint-detection
- object-detection
tags:
- spacecraft
- pose-estimation
- benchmark
ASTRA: Autonomous Spacecraft Tracking and Rendezvous Archive
Anonymous review copy for our NeurIPS 2026 Datasets and Benchmarks Track submission.
ASTRA is a large-scale multi-satellite benchmark for 6-DoF spacecraft pose estimation, covering twenty NASA-referenced spacecraft rendered in Unreal Engine 5 and paired with FRESCO and NVIDIA COSMOS real-style variants. See the accompanying paper for full details.
Contents
This anonymous release contains the test split of ASTRA, provided to support reviewer verification of the annotation schema and visual quality. The complete dataset, including all training and validation splits, will be made publicly available upon acceptance.
| File | Size |
|---|---|
testdata_20_spacecraft.tar.gz |
28.1 GB |
License
- Generated images and annotations: CC-BY-4.0
- Underlying CAD assets: NASA 3D Resources (NASA Public Domain)