RESIDE
A Benchmark for Single Image Dehazing
News :
> This page is no longer updated. Please check RESIDE-Standard and RESIDE- β for details.
>We have updated ITS, SOTS dataset ( Baidu Yun Link) and HSTS(both Dropbox and Baidu Yun Link). Please feel free to utilize it.
>We have updated SOTS dataset (both Dropbox and Baidu Yun Link). Please feel free to utilize it.
Abstract
We present a comprehensive study and evaluation of existing single image dehazing algorithms, using a new large-scale benchmark consisting of both synthetic and real-world hazy images, called REalistic Single Image DEhazing (RESIDE). RESIDE highlights diverse data sources and image contents, and is divided into five subsets, each serving different training or evaluation purposes. We further provide a rich variety of criteria for dehazing algorithm evaluation, ranging from full-reference metrics, to no-reference metrics, to subjective evaluation and the novel task-driven evaluation. Experiments on RESIDE shed light on the comparisons and limitations of state-of-the-art dehazing algorithms, and suggest promising future directions.
Overview
Dataset Examples:
Paper
RESIDE: A Benchmark for Single Image Dehazing [arXiv version]
Bibtex
@article{li2019benchmarking,
title={Benchmarking Single-Image Dehazing and Beyond},
author={Li, Boyi and Ren, Wenqi and Fu, Dengpan and Tao, Dacheng and Feng, Dan and Zeng, Wenjun and Wang, Zhangyang},
journal={IEEE Transactions on Image Processing},
volume={28},
number={1},
pages={492--505},
year={2019},
publisher={IEEE}
}
Download RESIDE Dataset
ITS (Indoor Training Set):
(Dropbox): http://t.cn/RHjBQIV
(Baidu Yun):https://pan.baidu.com/s/16rm4zUF8uVRs3Ux5T9CMMA Passward: tqyh
OTS (Outdoor Training Set):
(Dropbox): http://t.cn/RQXyZFI
(Baidu Yun): https://pan.baidu.com/s/1c2rW4hi Passward: 5vss
SOTS (Synthetic Objective Testing Set):
(Dropbox): http://t.cn/RQ34zUi
(Baidu Yun): https://pan.baidu.com/share/init?surl=SSVzR058DX5ar5WL5oBTLg Passward: s6tu
RTTS (Real-world Task-Driven Testing Set):
(Dropbox) :http://t.cn/RHP3eXg
(Baidu Yun) :https://pan.baidu.com/s/1nuJOdjr Passward: n3v8
HSTS (Hybrid Subjective Testing Set):
(Dropbox) :http://tinyurl.com/y7keuhvx
(Baidu Yun):https://pan.baidu.com/s/1cl1exWnaFXe3T5-Hr7TJIg Passward: vzeq
Unannotated Real-world Hazy Images:
(Dropbox) :http://t.cn/RHVjLXp
(Baidu Yun) : https://pan.baidu.com/s/1i5GiZsx Password:vt99
Note: For Baidu Yun Link, you might need to refresh again or copy the link to a new webpage if it fails to redirect.
请注意:如果您无法直接点击百度云链接访问,请开启一个新的网页并复制链接到地址栏。
How to Use RTTS
RTTS is in the same format as VOC 2007, organized as follows:
-- HazeDetection
-- Annotations
-- *** .xml
-- ImageSets
-- Main
-- test.txt
-- JPEGImages
-- *** .png
Here is a demo script to visualize the annotations: visualize.py, which shows how to get access to the annotated bounding boxes.
How to Evaluate PSNR, SSIM
To standardize the evaluation of PSNR and SSIM, here is our calculation code in RESIDE paper
(Both PSNR and SSIM from Matlab 2013)