SpaceSense-Bench: 136 Satellite Models for Multi-Modal Spacecraft Perception

SpaceSense-Bench is a high-fidelity simulation-based dataset for spacecraft component-level semantic understanding. It contains 136 satellite models sourced from NASA and ESA 3D models, with synchronized multi-modal data including RGB, Depth, and LiDAR point clouds. The dataset was created by Alvin16 and last updated on Hugging Face in April 2026.

Use Cases

• Train spacecraft component segmentation models based on the multi-modal RGB and depth data.
• Develop pose estimation algorithms for satellites based on the synchronized LiDAR point clouds.
• Benchmark multi-modal fusion techniques for space robotics using the provided RGB, depth, and point cloud modalities.
• Simulate and test on-orbit servicing scenarios using the 136 distinct satellite models.

Strengths

• Includes 136 distinct satellite models, providing a diverse set of spacecraft geometries.
• Sourced from authoritative 3D models provided by NASA and ESA.
• Provides synchronized multi-modal data (RGB, Depth, LiDAR), which is relatively rare for space applications.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and total dataset size are unknown, which may limit suitability assessment.
• Data is simulation-based, which may not fully capture the noise and conditions of real-space imagery.

Provenance

Source

Hugging Face dataset by Alvin16, with satellite models from NASA/ESA 3D models.

Collection Method

High-fidelity simulation-based generation.

Time Range

null

Freshness

Last updated 2026-04-02 14:03:35; freshness should be verified.

Geography

null