💥 Image De-Blurring Application 💥

⚡ TECH STACK USED

⚡ ABOUT THE APPLICATION

Face deblurring operation is the task of estimating a clear image from its degraded blur image and recovering the sharp contents and textures. The aim of face deblurring is to restore clear images with more explicit structure and facial details. The face deblurring problem has attracted considerable attention due to its wide range of applications.

Blurred Image	Processing	Clear Image

⚡ MODELS USED

Models Used	Accuracy
Autoencoder Model	89.80%
De-Blur CNN Model	78.20%

⚡ WHAT I HAVE DONE

1. Training an End-to-End model for deblurring of images (CelebA) following the work in CNN For Direct Text Deblurring, using Keras.

   • The first layer filter size is adjusted to be approximately equal to the blur kernel size. Pre-Trained model with weights and some images from test set are uploaded.

2. Importing Necessary Packages

3. Loading Images Dataset in the model

   • Only showing a small set of images from the local test set we generated.

4. Loading input blurred and clear images

5. Defining CNN Model for Training Model

   • The model has been trained on a much larger dataset of CelebA images.
   • Loaded the weight file autoencoder_weights.keras

6. Deblurred Faces

   • Deblurred images as output

⚡ RESULTS

⚡ De Blur CNN Model

⚡ Auto encoder CNN Model

⚡ Preview of Website

⚡ Predictions Done by Website

Show some  ❤️  by  🌟  this repository!