Image Deonising with Autoencoders project introduces a novel approach to denoise noisy images with Autoencoders. Here we make use of the Convolutional Autoencoders to build and train a Deep Neural Network which learns to remove noise from an image.
The following libraries are used for the project:
opencv-python
numpy
scikit-learn
matplotlib
tensorflow
In contrast to a classification problem, an autoencoder generates images. Hence we require only a very small dataset as this falls under a regression problem. We train on 200 samples and validate on 100 samples.
Go to the folder model and run the file inference.py
We have a novel approach with a simple architecture which learns to model gaussian noise.