Training data for Uni-RS: A Spatially Faithful Unified Understanding and Generation Model for Remote Sensing (IJCAI 2026).
Dataset Card
Training data for Uni-RS: A Spatially Faithful Unified Understanding and Generation Model for Remote Sensing (IJCAI 2026). [[Paper]](https://www.ijcai.org/proceedings/2026/231) [[Code]](https://github.com/zwy-Giser/Uni-RS) Unzip the three archives into one directory: One JSON object per line. Rows are aligned by index with the RSICap and VRSBench caption datasets. A plan contains SCENE (global scene), OBJ (object instances with a coarse position on a 3x3 grid: TopLeft, TopCenter,..., BottomRight, plus optional count and attribute) and REL (relations between objects) lines: One JSON object per line. Rows are aligned by index with the RSICap and VRSBench caption datasets.…
Excerpt from the card by WEIYU ZHANG, licensed cc-by-nc-4.0.
Details
- Repository
- Weiyu-Giser/RS-Spatial
- Publisher
- WEIYU ZHANG
- Task category
- Text to image
- Tags
- remote-sensing, spatial-relations, text-to-image
- Size category
- 10K<n<100K
- Languages
- en
- Revision
- 811ca079dd9e7b036f9021f632a0a469a728d876
- Last updated
- 2026-10-13
Files
5 files, 12.6 GB in total.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| layout_plans.zip | Data | 2.8 MB | 5c2b21825008 |
| rotation_pairs.zip | Data | 12.6 GB | b77d89f647ef |
| spatial_relations.zip | Data | 2.1 MB | 47f5b2f476d9 |
| README.md | Documentation | 4.5 KB | — |
| .gitattributes | Repository | 2.5 KB | — |
License and Download
- License
- cc-by-nc-4.0
- Access
- No access gate
Released by WEIYU ZHANG through its official repository on Hugging Face. Read the license.