This repository implements an image super-resolution algorithm based on the SRGAN architecture. The goal of this project is to enhance the resolution of low-resolution images by a factor of 4, resulting in visually sharper and more detailed images.
The DIV2K_HR dataset was used to train our SRGAN (Super-Resolution Generative Adversarial Network) model. DIV2K_HR is a popular dataset widely used in the field of image super-resolution research. It consists of high-resolution images collected from various sources, such as the internet, professional photography, and digital cameras. The dataset contains a diverse range of subjects, including natural landscapes, objects, animals, and people. To download the dataset, move to the project directory and run the following:
!wget http://data.vision.ee.ethz.ch/cvl/DIV2K/DIV2K_train_HR.zip
!unzip DIV2K_train_HR.zip
Create a conda environment
conda create -n super_res python=3.9.13
conda activate super_res
Change directory to project folder and install requirements
cd super_resolution
pip install -r requirements.txt
To train the model
python train.py
To inference on the model
streamlit run app.py