deblurgan-v2

## Use Case and High-Level Description

DeblurGAN-v2 is a generative adversarial network (GAN) for single image motion deblurring. This model is based on a relativistic conditional GAN with a double-scale discriminator. For details about architecture of model, check out the paper. Model used MobileNet as backbone and was trained on GoPro, DVD, NFS datasets. For details about implementation of model, check out the DeblurGAN-v2: Deblurring (Orders-of-Magnitude) Faster and Better repository.

## Specification

Metric Value
Type Image Processing
GFLOPs 80.8919
MParams 2.1083
Source framework PyTorch*

## Accuracy

Model was tested on GoPro test dataset.

Metric Original model Converted model
PSNR 28.25Db 28.24Db
SSIM 0.97 0.97

## Input

### Original Model

Image, name - blur_image, shape - 1, 3, 736, 1312, format is B, C, H, W, where:

• B - batch size
• C - channel
• H - height
• W - width

Channel order is RGB. Mean values - [127.5, 127.5, 127.5], scale values - [127.5, 127.5, 127.5].

### Converted Model

Image, name - blur_image, shape - 1, 3, 736, 1312, format is B, C, H, W, where:

• B - batch size
• C - channel
• H - height
• W - width

Channel order is BGR.

## Output

### Original Model

Deblurred image, name - deblur_image, shape - 1, 3, 736, 1312, output data format is B, C, H, W, where:

• B - batch size
• C - channel
• H - height
• W - width

Channel order is RGB.

### Converted Model

Deblurred image, name - deblur_image, shape - 1, 3, 736, 1312, output data format is B, C, H, W, where:

• B - batch size
• C - channel
• H - height
• W - width

Channel order is BGR.

You can download models and if necessary convert them into Inference Engine format using the Model Downloader and other automation tools as shown in the examples below.

An example of using the Model Converter:

## Legal Information

Copyright (c) 2017, Jun-Yan Zhu and Taesung Park
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice, this
list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the documentation
and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
For pix2pix software
Copyright (c) 2016, Phillip Isola and Jun-Yan Zhu
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice, this
list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the documentation
and/or other materials provided with the distribution.