A tensorflow implement of the TIP2017 paper Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising

• BSD68 Average Result

The average PSNR(dB) results of different methods on the BSD68 dataset.

• Set12 Average Result

For the dataset and denoised images, please download here

• Install docker support

You may do it like this(ubuntu):

$ sudo apt-get install -y curl
$ curl -sSL https://get.docker.com/ | sh
$ sudo usermod -aG docker ${USER}

• Install nvidia-docker support(to make your GPU available to docker containers)

You may do it like this(ubuntu):

$ wget -P /tmp https://github.com/NVIDIA/nvidia-docker/releases/download/v1.0.1/nvidia-docker_1.0.1-1_amd64.deb
$ sudo dpkg -i /tmp/nvidia-docker*.deb && rm /tmp/nvidia-docker*.deb

• Pull dncnn image and start a container

$ docker pull wenbodut/dncnn
$ ./rundocker.sh

Then you could train the model.

You should make sure the following environment is contented

tensorflow == 1.4.1
numpy

$ ./oneKeyToDenoise.sh
(need docker support)

Then you could find the noisy Set12 images and denoised images in test folder. Have fun!

$ python generate_patches.py
$ python main.py
(note: You can add command line arguments according to the source code, for example
    $ python main.py --batch_size 64 )

For the provided model, it took about 4 hours in GTX 1080TI.

Here is my training loss:

Note: This loss figure isn't suitable for this trained model any more, but I don't want to update the figure 🌚

$ python main.py --phase test

• Fix bug #13. (bug #13 fixed, thanks to @sdlpkxd)
• Clean source code. For instance, merge similar functions(e.g., 'load_images 'and 'load_image' in utils.py).
• Add one-key denoising, with the help of docker.
• Compare with original DnCNN.
• Replace tf.nn with tf.layer.
• Replace PIL with OpenCV.
• Try tf.dataset API to speed up training process.
• Train a noise level blind model.

• @lizhiyuanUSTC
• @husqin
• @sdlpkxd
• and so on ...