Image Reconstruction Using CNN Based Upsampling

This project involves a Python program that performs downsampling of images, converts them to the YUV color space, and then upsamples them using a trained Convolutional Neural Network (CNN) model.

Prerequisites

To install all required packages run the following command:
pip install -r requirements.txt

Usage

1. Define the input image at the top of the main.py file
2. Run python3 main.py

Demo

             

The original (left) is converted to the YUV and donwsampled (middle). The Y channel is downsampled by a factor of 2 as it contains key image data while the U and V channels are downsampled by a factor of 4. The individual channels are then upsampled, using their corresponding CNN trained models. The upsampled channels are then concatenated to form the reconstructed image (right)

License:

This project is licensed under the MIT