gdagil / SwinIR-docker

SwinIR transformer in container

SwinIR-docker

Simple image to upscaling images and videos in .mp4 format by SwinIR model

Fast way to use it with default settings

Place the files to be processed in data/raw_data directory and run the script with the following command lines

cd SwinIR-docker 
sudo docker-compose up

RAW	UPSCALED

Usung the image

So, if you want only build the image and run it with SwinIR settings you can need use environment variables

• TASK_NAME
• SCALE
• MODEL
• MODEL_PATH
• TR_PS - training_patch_size
• NOISE
• JPEG
• TILE
• TILE_OVERLAP

ID	TASK_NAME	MODEL	SCALE	TR_PS	NOISE	JPEG	TILE	MODEL_PATH
1	real_sr	large_model	4					model_zoo/swinir/003_realSR_BSRGAN_DFOWMFC_s64w8_SwinIR-L_x4_GAN.pth
2			4				400	model_zoo/swinir/003_realSR_BSRGAN_DFO_s64w8_SwinIR-M_x4_GAN.pth
3	classical_sr		2	48				model_zoo/swinir/001_classicalSR_DIV2K_s48w8_SwinIR-M_x2.pth
4			3	48				model_zoo/swinir/001_classicalSR_DIV2K_s48w8_SwinIR-M_x3.pth
5			4	48				model_zoo/swinir/001_classicalSR_DIV2K_s48w8_SwinIR-M_x4.pth
6			8	48				model_zoo/swinir/001_classicalSR_DIV2K_s48w8_SwinIR-M_x8.pth
7			2	64				model_zoo/swinir/001_classicalSR_DIV2K_s64w8_SwinIR-M_x2.pth
8			3	64				model_zoo/swinir/001_classicalSR_DIV2K_s64w8_SwinIR-M_x3.pth
9			4	64				model_zoo/swinir/001_classicalSR_DIV2K_s64w8_SwinIR-M_x4.pth
10			8	64				model_zoo/swinir/001_classicalSR_DIV2K_s64w8_SwinIR-M_x8.pth
11	lightweight_sr		2	48				model_zoo/swinir/002_lightweightSR_DIV2K_s64w8_SwinIR-S_x2.pth
12			3	48				model_zoo/swinir/002_lightweightSR_DIV2K_s64w8_SwinIR-S_x3.pth
13			4	48				model_zoo/swinir/002_lightweightSR_DIV2K_s64w8_SwinIR-S_x4.pth
14	gray_dn				15			model_zoo/swinir/004_grayDN_DFWB_s128w8_SwinIR-M_noise15.pth
15					25			model_zoo/swinir/004_grayDN_DFWB_s128w8_SwinIR-M_noise25.pth
16					50			model_zoo/swinir/004_grayDN_DFWB_s128w8_SwinIR-M_noise50.pth
17	color_dn				15			model_zoo/swinir/005_colorDN_DFWB_s128w8_SwinIR-M_noise15.pth
18					25			model_zoo/swinir/005_colorDN_DFWB_s128w8_SwinIR-M_noise25.pth
19					50			model_zoo/swinir/005_colorDN_DFWB_s128w8_SwinIR-M_noise50.pth
20	jpeg_car					10		model_zoo/swinir/006_CAR_DFWB_s126w7_SwinIR-M_jpeg10.pth
21						20		model_zoo/swinir/006_CAR_DFWB_s126w7_SwinIR-M_jpeg20.pth
22						30		model_zoo/swinir/006_CAR_DFWB_s126w7_SwinIR-M_jpeg30.pth
23						40		model_zoo/swinir/006_CAR_DFWB_s126w7_SwinIR-M_jpeg40.pth

For eample settings №1:

docker build .
docker run --rm \
  -e TASK_NAME=real_sr \
  -e SCALE=4 -e MODEL=large_model \
  -e MODEL_PATH=model_zoo/swinir/003_realSR_BSRGAN_DFOWMFC_s64w8_SwinIR-L_x4_GAN.pth \
  danil2286/swin.ir:latest

Citation

@article{liang2021swinir,
  title={SwinIR: Image Restoration Using Swin Transformer},
  author={Liang, Jingyun and Cao, Jiezhang and Sun, Guolei and Zhang, Kai and Van Gool, Luc and Timofte, Radu},
  journal={arXiv preprint arXiv:2108.10257},
  year={2021}
}