jdopensource/JoyAI-Image-SpatialEdit

---
license: apache-2.0
---

# SpatialEdit-500K

SpatialEdit-500K is a synthetic training dataset for fine-grained image spatial editing. It is built for learning geometry-aware edits such as object moving, object rotation, and camera viewpoint change.

The dataset was introduced in the paper [SpatialEdit: Benchmarking Fine-Grained Image Spatial Editing](https://huggingface.co/papers/2604.04911). It is generated with a controllable rendering pipeline to provide structured spatial transformations at scale.

This dataset is used in [jd-opensource/JoyAI-Image](https://github.com/jd-opensource/JoyAI-Image)

## Project Resources

- **GitHub Repository:** [EasonXiao-888/SpatialEdit](https://github.com/EasonXiao-888/SpatialEdit)

## Highlights

- **Large-scale synthetic data:** 500,000 samples for spatially grounded image editing.
- **Comprehensive transformations:** Covers both object-centric (moving, rotation) and camera-centric transformations.
- **High fidelity:** Generated with a controllable Blender pipeline rendering objects across diverse backgrounds with systematic camera trajectories.
- **Precise labels:** Provides precise ground-truth transformations for spatial manipulation tasks.

## Citation

```bibtex
@article{xiao2026spatialedit,
title = {SpatialEdit: Benchmarking Fine-Grained Image Spatial Editing},
author = {Xiao, Yicheng and Zhang, Wenhu and Song, Lin and Chen, Yukang and Li, Wenbo and Jiang, Nan and Ren, Tianhe and Lin, Haokun and Huang, Wei and Huang, Haoyang and Li, Xiu and Duan, Nan and Qi, Xiaojuan},
journal = {arXiv preprint arXiv:2604.04911},
year = {2026}
}
```