Skip to content

Latest commit

 

History

History

nafnet

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
LICENSE
LICENSE
 
 
README.md
README.md
 
 
blur_img.png
blur_img.png
 
 
deblur_output.png
deblur_output.png
 
 
nafnet.py
nafnet.py
 
 
noise_output.png
noise_output.png
 
 
noisy.png
noisy.png
 
 

README.md

NAFNet

denoising

Input

Input

(Image from https://github.com/megvii-research/NAFNet/blob/main/demo/denoise_img.png)

Ailia input shape : (1, 3, IMAGE_HEIGHT, IMAGE_WIDTH)

Output

Output

Ailia output shape : (1, 3, IMAGE_HEIGHT, IMAGE_WIDTH)

Usage

Automatically downloads the onnx and prototxt files when running. It is necessary to be connected to the Internet while downloading.

For the sample image with twice the resolution,

$ python3 nafnet.py

If you want to specify the input image, put the image path after the --input option.
You can use --savepath option to change the name of the output file to save.

$ python3 nafnet.py --input IMAGE_PATH --savepath SAVE_IMAGE_PATH

By adding the --arch option, you can specify architecture type which is selected from 'Baseline-SIDD-width32', 'NAFNet-SIDD-width64', 'Baseline-SIDD-width64' ,'NAFNet-SIDD-width32' ,(default is NAFNet-SIDD-width32)

$ python3 nafnet.py --arch NAFNet-SIDD-width32

By adding the --video option, you can input the video.
If you pass 0 as an argument to VIDEO_PATH, you can use the webcam input instead of the video file.

$ python3 nafnet.py --video VIDEO_PATH

deblurring

Input

Input

(Image from https://github.com/megvii-research/NAFNet/blob/main/demo/denoise_img.png)

Ailia input shape : (1, 3, IMAGE_HEIGHT, IMAGE_WIDTH)

Output

Output

Ailia output shape : (1, 3, IMAGE_HEIGHT, IMAGE_WIDTH)

Usage

Automatically downloads the onnx and prototxt files when running. It is necessary to be connected to the Internet while downloading.

For the sample image with twice the resolution,

$ python3 nafnet.py --arch NAFNet-REDS-width64

If you want to specify the input image, put the image path after the --input option.
You can use --savepath option to change the name of the output file to save.

$ python3 nafnet.py --input IMAGE_PATH --savepath SAVE_IMAGE_PATH --arch NAFNet-REDS-width64

By adding the --arch option, you can specify architecture type which is selected from 'Baseline-GoPro-width32' ,'NAFNet-GoPro-width32', 'NAFNet-REDS-width64', 'Baseline-GoPro-width64','NAFNet-GoPro-width64'

$ python3 nafnet.py --arch NAFNet-REDS-width64

By adding the --video option, you can input the video.
If you pass 0 as an argument to VIDEO_PATH, you can use the webcam input instead of the video file.

$ python3 nafnet.py --video VIDEO_PATH --arch NAFNet-REDS-width64

Reference

NAFNet

Framework

Pytorch

Model Format

ONNX opset=11

Netron

Baseline-GoPro-width32.onnx.prototxt

Baseline-GoPro-width64.onnx.prototxt

Baseline-SIDD-width32.onnx.prototxt

Baseline-SIDD-width64.onnx.prototxt

NAFNet-GoPro-width32.onnx.prototxt

NAFNet-GoPro-width64.onnx.prototxt

NAFNet-REDS-width64.onnx.prototxt

NAFNet-SIDD-width32.onnx.prototxt

NAFNet-SIDD-width64.onnx.prototxt