nushrat-tarmin-meem/syra-synthesized_rain_dataset

Synthesized Rain dataset

SyRaGAN, NeuroComputing 2022

Pytorch implementation

Update completion

SyRa : Synthesized rain image for deraining algorithms Paper

Jaewoong Choi, Daeha Kim, Sanghyuk Lee, Byung Cheol Song

On this repository, SyRaGAN's code and instructions for synthesizing rain images are explained.

Requirements

Install the dependencies:

bash
conda create -n SyRaGAN python=3.6.7
conda activate SyRaGAN
conda install -y pytorch=1.4.0 torchvision=0.5.0 cudatoolkit=10.0 -c pytorch
conda install x264=='1!152.20180717' ffmpeg=4.0.2 -c conda-forge
pip install opencv-python==4.1.2.30 ffmpeg-python==0.2.0 scikit-image==0.16.2
pip install pillow==7.0.0 scipy==1.2.1 tqdm==4.43.0 munch==2.5.0
pip install tqdm

Pretrained model

Click to download pretrained SyRa-GAN

SyRa dataset

Click to download SyRa
: Trainset - 10K clear image and 50K synthesized rain image , Testset - 1K clear image and 5K synthesized rain image

Click to download SyRa-HQ
: Trainset - 1K clear image and 5K synthesized rain image , Testset - 100 clear image and 500 synthesized rain image

Training dataset

As training data, Rain100L [1], Rain100H [1], Rain800 [2], Rain1200 [3], Rain1400 [4], and SPA-data [5] were used. The training image is used by concating each clear image and rain image.

Training SyRaGAN

Divide your training images into the following locations : ./data/rains/train/A ./data/rains/train/B

Example of training image :

Run

python main.py --img_size 256 --mode train --checkpoint_dir expr/checkpopint/SyRa --resume_iter 0 --gpu 0

Synthesizing rain image

Put clear images in the following location. ./asset/folder_of_your_data

Put checkpoint file in the following location. ./expr/checkpoint/SyRa

Run

python main.py --img_size 256 --mode syn --checkpoint_dir expr/checkpoint/SyRa --out_dir expr/result --data folder_of_your_data --resume_iter 100000

5 syntheiszed rain images will be created for each clear image in ./expr/result

References

[1] Yang, Wenhan, et al. "Deep joint rain detection and removal from a single image." Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.

[2] Zhang, He, Vishwanath Sindagi, and Vishal M. Patel. "Image de-raining using a conditional generative adversarial network." IEEE transactions on circuits and systems for video technology 30.11 (2019): 3943-3956.

[3] Zhang, He, and Vishal M. Patel. "Density-aware single image de-raining using a multi-stream dense network." Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.

[4] Fu, Xueyang, et al. "Removing rain from single images via a deep detail network." Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017.

[5] Wang, Tianyu, et al. "Spatial attentive single-image deraining with a high quality real rain dataset." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019.