Model Description

Keras Implementation of Convolutional autoencoder for image denoising

This repo contains the trained model of Convolutional autoencoder for image denoising on MNIST Dataset mixed with random noise.

Spaces Link:- https://huggingface.co/spaces/keras-io/conv_autoencoder

Keras Example Link:- https://keras.io/examples/vision/autoencoder/

Intended uses & limitations

• The trained model can be used to remove noise from any grayscale image.
• Since this model is trained on MNIST Data added with random noise, so this model can be used only for images with shape 28 * 28.

Training and evaluation data

• Original mnist train & test dataset were loaded from tensorflow datasets.
• Then Some noise was added to train & test images.
• Noisy images were used as input images and original clean images were used as output images for training.

Training procedure

Training hyperparameter

The following hyperparameters were used during training:

• optimizer: 'adam'
• loss: 'binary_crossentropy'
• epochs: 100
• batch_size: 128
• ReLU was used as activation function in all layers except last layer where Sigmoid was used as activation function.

Model Plot

<details> <summary>View Model Plot</summary>

</details>