Deep convolutional autoencoder and variational autoencoder.
Denoise-VAE-keras.ipynb use keras with tensorflow backend, it is a variational autoencoder model.
Denoise-AE-tf.ipynb use tensorflow, it is an autoencoder model.
The six pictures in first row are the original pictures.
The second are the noised pictures with random noises.
The third are the results of the Denoise-AE-tf.ipynb
The forth are the results of median filtering, which is relized by the scipy.signal.
git clone https://github.com/block98k/Denoise-VAE.git
cd ./Denoise-VAE
jupyter notebook Denoise-AE-tf.ipynb
tensorflow 1.0 or above
numpy
scipy
jupyter
This is a project I used to practice autoencoder.