% =============================================================== The dataset in this package provides the real-world noisy images as described in the following paper: Jun Xu, Hui Li, Zhetong Liang, David Zhang, and Lei Zhang Real-world Noisy Image Denoising: A New Benchmark https://arxiv.org/abs/1804.02603, 2018. Please cite the paper if you are using this dataset in your research. Please see the file License.txt for the license governing this code. Version: 1.0 (04/07/2018), see ChangeLog.txt Contact: Jun Xu <csjunxu@comp.polyu.edu.hk/nankaimathxujun@gmail.com> % =============================================================== Note --------------- For more training and testing needs, please refer to https://pan.baidu.com/s/1539s_gNN8zvDYxG-DbiDuA (code: gy9b) for more shots of the real-world noisy images. Overview --------------- This dataset contains 37 different scenes captured by 5 cameras from the 3 leading brands of cameras: 1) Canon EOS (5D Mark II, 80D, 600D); 2) Nikon (D800); 3) Sony (A7 II). We crop 100 regions of 512X512 from these 37 scenes1: The *Real.JPG are noisy images; The *mean.JPG are "ground truth" images. Dataset Details --------------- Camera 1: Canon EOS 5D Mark II Image Name Size Aperture Shutter Speed ISO Value Canon5D2_bicyc 2784 x 1856 f/5 1/160s 6400 Canon5D2_chair 2784 x 1856 f/5 1/160s 3200 Canon5D2_circu 2784 x 1856 f/5 1/160s 6400 Canon5D2_desk 2784 x 1856 f/5 1/160s 6400 Canon5D2_fruit 2784 x 1856 f/5 1/200s 3200 Canon5D2_mouse 2784 x 1856 f/5 1/160s 3200 Canon5D2_plug 2784 x 1856 f/5 1/160s 3200 Canon5D2_recie 2784 x 1856 f/5 1/160s 6400 Canon5D2_robot 2784 x 1856 f/5 1/160s 3200 Canon5D2_toy 2784 x 1856 f/5 1/200s 3200 Camera 2: Canon EOS 80D Image Name Size Aperture Shutter Speed ISO Value Canon80D_ball 2976 x 1680 f/8 1/8s 3200 Canon80D_compr 2976 x 1680 f/8 1/8s 6400 Canon80D_corne 2976 x 1680 f/8 1/8s 1600 Canon80D_GO 2976 x 1680 f/8 1/8s 800 Canon80D_print 2976 x 1680 f/8 1/8s 12800 Camera 3: Canon EOS 600D Image Name Size Aperture Shutter Speed ISO Value Canon600_book 5184 x 3456 f/4.5 1/125s 1600 Canon600_toy 5184 x 3456 f/4.5 1/125s 1600 Canon600_water 5184 x 3456 f/3.5 1/125s 1600 Camera 4: NIKON D800 Image Name Size Aperture Shutter Speed ISO Value Nikon800_bulle 3680 x 2456 f/8 1/100s 6400 Nikon800_class 3680 x 2456 f/4.5 1/160s 1600 Nikon800_desch 3680 x 2456 f/11 1/160s 3200 Nikon800_desk 3680 x 2456 f/4.5 1/160s 3200 Nikon800_door 3680 x 2456 f/5.6 1/160s 6400 Nikon800_flowe 3680 x 2456 f/5 1/100s 4000 Nikon800_map 3680 x 2456 f/5 1/100s 4000 Nikon800_photo 3680 x 2456 f/8 1/125s 6400 Nikon800_plant 3680 x 2456 f/6.3 1/125s 5000 Nikon800_plaso 3680 x 2456 f/10 1/100s 6400 Nikon800_stair 3680 x 2456 f/5 1/125s 6400 Nikon800_wall 3680 x 2456 f/5 1/100s 6400 Camera 5: SonyA7II ILCE-7M2 Image Name Size Aperture Shutter Speed ISO Value SonyA7II_book 3008 x 1688 f/4.5 1/125s 1600 SonyA7II_class 3008 x 1688 f/3.5 1/200s 1600 SonyA7II_compu 3008 x 1688 f/3.5 1/500s 3200 SonyA7II_door 3008 x 1688 f/4 1/200s 3200 SonyA7II_plant 3008 x 1688 f/4.5 1/125s 3200 SonyA7II_toy 3008 x 1688 f/4.5 1/125s 1600 SonyA7II_water 3008 x 1688 f/4.5 1/125s 6400 Other Datasets --------------- CC: 15 cropped real-world noisy images from CC [1]. This dataset can be found at http://snam.ml/research/ccnoise The smaller 15 cropped images can be found on in the directory ''Real_ccnoise_denoised_part'' of https://github.com/csjunxu/MCWNNM_ICCV2017 The *real.png are noisy images; The *mean.png are "ground truth" images; The *ours.png are images denoised by CC. DND_2017: 1000 cropped real-world noisy images from DND [2]. Please download the dataset from https://noise.visinf.tu-darmstadt.de/ and put the files in "DND_2017" directory accordingly. [1] Seonghyeon Nam*, Youngbae Hwang*, Yasuyuki Matsushita, Seon Joo Kim. A Holistic Approach to Cross-Channel Image Noise Modeling and its Application to Image Denoising. CVPR, 2016. [2] Tobias Pl?tz and Stefan Roth. Benchmarking Denoising Algorithms with Real Photographs. CVPR, 2017. Dependency ------------ This dataset does not depend on any external dataset. Contact ------------ If you have questions, problems with the code, or find a bug, please let us know. Contact Jun Xu at csjunxu@comp.polyu.edu.hk or nankaimathxujun@gmail.com