---
title: NH-HAZE
canonical_url: "https://www.modelscope.cn/datasets/OpenDataLab/NH-HAZE"
md_url: "https://www.modelscope.cn/datasets/OpenDataLab/NH-HAZE.md"
repository: OpenDataLab/NH-HAZE
last_updated: 2025-02-27
license: "[Unknown]"
storage_size: "603 MB"
domain:
  - publishDate
  - paperUrl
  - publishUrl
  - mediaTypes
  - displayName
  - taskTypes
  - publisher
tasks:
  - 2020
  - "https://data.vision.ee.ethz.ch/cvl/ntire20/nh-haze/files/NH_HAZE_IEEE.pdf"
  - "https://homes.cs.washington.edu/~msap/social-bias-frames/"
  - Image
  - NH-HAZE
  - "Single Image Dehazing"
  - "University of Girona"
downloads: 2000
stars: 0
---

# NH-HAZE

> NH-HAZE - OpenDataLab 在 ModelScope 开源的数据集。displayName: NH-HAZE license: Unknown mediaTypes: Image paperUrl: https://data.vision.ee.ethz.ch/cvl/ntire20/nh-haze/files/NHHAZEIEEE.pdf publishDate: "2020" publishUrl:…

OpenDataLab/NH-HAZE 是 ModelScope 魔搭社区上的2020、https://data.vision.ee.ethz.ch/cvl/ntire20/nh-haze/files/NH_HAZE_IEEE.pdf、https://homes.cs.washington.edu/~msap/social-bias-frames/数据集，涉及 publishDate、paperUrl、publishUrl 领域，存储大小 603 MB，采用 [Unknown] 许可。

- **Repository**: OpenDataLab/NH-HAZE
- **License**: [Unknown]
- **Tasks**: 2020, https://data.vision.ee.ethz.ch/cvl/ntire20/nh-haze/files/NH_HAZE_IEEE.pdf, https://homes.cs.washington.edu/~msap/social-bias-frames/, Image, NH-HAZE, Single Image Dehazing, University of Girona
- **Domain**: publishDate, paperUrl, publishUrl, mediaTypes, displayName, taskTypes, publisher
- **Storage size**: 603 MB
- **Downloads**: 2000
- **Stars**: 0
- **Last updated**: 2025-02-27

Source: https://www.modelscope.cn/datasets/OpenDataLab/NH-HAZE

---

displayName: NH-HAZE
license:
- Unknown
mediaTypes:
- Image
paperUrl: https://data.vision.ee.ethz.ch/cvl/ntire20/nh-haze/files/NH_HAZE_IEEE.pdf
publishDate: "2020"
publishUrl: https://homes.cs.washington.edu/~msap/social-bias-frames/
publisher:
- ETH Zurich
- Politehnica University of Timisoara
- University of Girona
tags:
- Tree
- Chair
taskTypes:
- Image Dehazing
- Single Image Dehazing

---
# 数据集介绍
  ## 简介
   图像去雾是近年来被广泛研究的病态问题。由于缺乏参考数据集，去雾方法的客观性能评估是主要障碍之一。虽然合成数据集已显示出重要的局限性，但最近引入的少数现实数据集假设整个场景的雾度均匀。由于在许多实际情况下，雾度不是均匀分布的，我们引入了 NH-HAZE，这是一个非均匀真实数据集，具有成对的真实雾度和相应的无雾度图像。这是第一个非同质图像去雾数据集，包含 55 个户外场景。使用模拟雾霾场景真实情况的专业雾度发生器在场景中引入非均匀雾度。此外，这项工作对使用 NH-HAZE 数据集评估的几种最先进的单图像去雾方法进行了客观评估。 
  ## 引文
  
```
"@inproceedings{ancuti2020nh,
title={NH-HAZE: An image dehazing benchmark with non-homogeneous hazy and haze-free images},
author={Ancuti, Codruta O and Ancuti, Cosmin and Timofte, Radu},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},
pages={444--445},
year={2020}
}"
```

  
## Download dataset
:modelscope-code[]{type="git"}