Upscale-App

Upscale-App is a web application that enhances the quality of images by improving their resolution. This project utilizes advanced deep learning models, namely EDSR_x4 (Enhanced Deep Super-Resolution) and LapSRN_x8 (Laplacian Pyramid Super-Resolution Network), to upscale images and provide users with high-resolution outputs. The app is built using Python for the backend, and HTML, JQuery, and Ajax for the frontend.

Features

• Image Upload: Users can upload their images to the application.
• Image Upscaling: The app enhances the image resolution using two advanced models:
  • EDSR_x4: A model designed to provide a 4x resolution enhancement.
  • LapSRN_x8: A model designed for 8x resolution enhancement.
• Ajax-Based Image Processing: Asynchronous requests ensure a smooth user experience without needing to reload the page.

Technologies Used

Backend:

• Python: The core language used for model implementation and image processing.
• OpenCV: For image manipulation and preprocessing.

Frontend:

• HTML: Used to create the structure of the web pages.
• JQuery: For dynamic interactions and triggering image upload and processing.
• Ajax: To send asynchronous requests to the backend for processing the images.

Models Used

1. EDSR_x4 (Enhanced Deep Super-Resolution)

EDSR is a state-of-the-art super-resolution model that enhances image quality by learning high-frequency details from low-resolution images. The x4 version of the model upscales images by a factor of 4, improving fine details and textures without introducing significant artifacts.

• Paper: https://arxiv.org/abs/1707.02921
• Features:
  • Better detail retention for sharp images.
  • Avoids unnecessary batch normalization, improving the final image quality.
  • Ideal for general use cases.

2. LapSRN_x8 (Laplacian Pyramid Super-Resolution Network)

LapSRN is a progressive super-resolution model that reconstructs high-resolution images using a pyramid framework. The x8 version upscales images by a factor of 8, allowing for more extreme upscaling while maintaining visual fidelity.

• Paper: https://arxiv.org/abs/1710.01992
• Features:
  • Multi-level pyramid structure for progressive image upscaling.
  • Efficient memory usage with better sharpness for extreme resolutions.
  • Ideal for images requiring large-scale upscaling.

Installation

1. Clone the repository:

   git clone https://github.com/fadingbeat/upscale-app.git

2. Navigate to the project directory:

   cd upscale-app

3. Create virtual environment:

   python -m venv venv

4. Activate virtual environment:

   venv\Scripts\activate

   or

   ./Scripts/activate.bat

5. Install the required dependencies:

   pip install -r requirements.txt

6. Run the application:

   python upscale.py

Usage

1. Open your browser and navigate to http://localhost:5000.
2. Upload an image using the file input.
3. Click on the "Upscale Image" button.
4. Download or view the enhanced image once processing is complete.

Note: The image processing could take up to 15 minutes if you're running on a low computational laptop.

Project Structure

upscale-app/
│
├── models/                 # Contains pre-trained model weights for EDSR_x4 and LapSRN_x8
├── static/                 # Static files like CSS and JS
├── templates/              # HTML templates for rendering web pages
├── upscale.py              # Main application script
├── requirements.txt        # Python dependencies
└── README.md               # Project documentation

Future Improvements

• Support for additional super-resolution models.
• Batch image processing.
• Integration of GPU support for faster image upscaling.
• Mobile-friendly UI enhancements.

---

Feel free to contribute to this project by submitting pull requests or reporting issues!