README.md

---
{}
---

<h1 align="center">
From Ideal to Real: Unified and Data-Efficient Dense Prediction for Real-World Scenarios
</h1>
<p align="center">
  <a href="https://xcltql666.github.io/DenseDiTProj/"><b>[🌐 Website]</b></a> •
  <a href="https://arxiv.org/abs/2506.20279"><b>[📜 Paper]</b></a> •
  <a href="https://huggingface.co/xcll/DenseDiT_Model"><b>[🤗 HF Models]</b></a> •  
  <a href="https://huggingface.co/datasets/xcll/DenseWorld_Benchmark"><b>[🤗 HF Dataset]</b></a> •  
</p>


<p align="center">
Repo for "<a href="https://arxiv.org/abs/2506.20279" target="_blank">From Ideal to Real: Unified and Data-Efficient Dense Prediction for Real-World Scenarios</a>"
</p>

![Visual](./visual_examples.jpg)

## DenseDiT Model
![DenseDiT](./densediT.jpg)

## DenseWorld Benchmark
![DenseWorld](./denseworld.jpg)

## Citation
If you find it helpful, please kindly cite the paper.
```
@misc{xia2025idealrealunifieddataefficient,
      title={From Ideal to Real: Unified and Data-Efficient Dense Prediction for Real-World Scenarios}, 
      author={Changliang Xia and Chengyou Jia and Zhuohang Dang and Minnan Luo},
      year={2025},
      eprint={2506.20279},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2506.20279}, 
    }
```