ColorSplash

A Flask web application that colorizes black and white images using DeOldify AI.

Features

• Upload black and white or grayscale images
• Automatically colorize images using AI
• View side-by-side comparison of original, grayscale, and colorized versions
• Simple and intuitive web interface

Installation

1. Clone this repository:

   git clone https://github.com/ahajin07/colorplash.git
   cd colorplash
   

2. Create a virtual environment and install dependencies:

   python -m venv venv
   
   # On Windows:
   venv\Scripts\activate
   
   # On macOS/Linux:
   source venv/bin/activate
   
   pip install -r requirements.txt
   

3. Set up DeOldify (the AI colorization model):

   git clone https://github.com/jantic/DeOldify.git
   cd DeOldify
   pip install -e .
   cd ..
   

4. Download the DeOldify model file:

   • Download the ColorizeArtistic_gen.pth file from this link
   • Place it in the models directory

Directory Structure

colorplash/
│
├── app.py                    # Main Flask application
├── colorizer_utils.py        # Utility for colorization
├── models/                   # Model storage
│   └── ColorizeArtistic_gen.pth  # DeOldify model file (needs to be downloaded)
├── static/                   # Static assets
│   └── uploads/              # Image upload storage
├── templates/                # HTML templates
│   └── index.html            # Main app page
├── DeOldify/                 # DeOldify submodule
└── requirements.txt          # Python dependencies

Usage

1. Run the application:

   python app.py
   

2. Open your browser and navigate to http://127.0.0.1:5000

3. Upload a black and white image and see the colorized result

How It Works

ColorSplash uses the DeOldify project, which implements a novel approach to image colorization using self-attention mechanisms in GANs (Generative Adversarial Networks). The app:

1. Accepts user-uploaded images
2. Converts them to grayscale for consistent input
3. Processes them through the DeOldify artistic model
4. Displays the original, grayscale, and colorized versions

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• This project relies on DeOldify by Jason Antic
• Built with Flask