InkSpire - Sketch to Image with GANs

InkSpire is a web application that leverages Generative Adversarial Networks (GANs) to convert sketches into colored images in real-time. This project combines an interactive frontend canvas with a Flask backend powered by a custom-trained GAN model, which was trained on the Anime Sketch Colorization Pair Dataset. It provides users with a seamless and dynamic sketch-to-image translation experience.

Contributors:

• Samrat Ray (IIT2023066)
• Ranjeet Kulkarni (IIT2023064)

Link to Models (Download the Entire Folder):

• https://drive.google.com/drive/folders/1P6_rF7ZAQ83np2US7JAE6TUR4mpTRBqL?usp=sharing

---

Key Features

• 🌟 Instant Image Transformation: Sketch your ideas and see them come to life with vibrant colors, powered by GANs.

• ✍️ Interactive Drawing Canvas: Draw freely and watch the AI generate a colorized version of your sketch in real-time.

• 🎨 AI-Enhanced Artwork: Add simple sketches, outlines, or doodles, and let InkSpire enhance them with AI-driven colorization.

• 🚀 Seamless User Experience: With an easy-to-use interface, InkSpire ensures that anyone—from beginners to professionals—can create AI-powered artwork effortlessly.

---

Requirements

Before starting, ensure you have the following installed:

System Requirements

• Python 3.x
• HTML, CSS, and JavaScript (for the frontend)

Python Libraries

To install the required Python libraries, run:

pip install -r requirements.txt

Frameworks

• Flask (for backend)
• TensorFlow

---

Installation

1. Clone the Repository

Clone the project to your local machine:

git clone https://github.com/your-username/inkspire.git
cd inkspire

2. Install Dependencies

Ensure all dependencies are installed:

pip install -r requirements.txt

3. Set Up the Flask Backend

Run the Flask server:

python app.py

This starts the backend server at http://localhost:5000.

4. Frontend Setup

The frontend uses an HTML5 canvas for drawing sketches:

1. Open index.html in your browser.
2. Draw sketches on the canvas.
3. View the real-time colorized results generated by the GAN model.

5. Running the App

With the Flask server running and the frontend loaded, you can:

• Start drawing on the canvas.
• View the colorized image generated in real-time.

---

How It Works

DataSet

Anime Sketch Colorization Pair Dataset

Frontend

• Users draw on an interactive HTML5 canvas.
• JavaScript sends the canvas data to the Flask backend every second.

Backend (Flask)

• Processes the incoming sketches.
• Sends the sketches to the custom-trained GAN model for colorization.

GAN Model

• A custom-trained GAN model generates colored images from sketches.
• The model was trained on the Anime Sketch Colorization Pair Dataset.
• Results are sent back to the frontend for display.

---

Technologies Used

Core Technologies

• Generative Adversarial Networks (GANs): For sketch-to-image translation.
• Flask: Python web framework for the backend.
• TensorFlow: For implementing and running the GAN model.
• HTML5 Canvas: Interactive drawing area for users.
• JavaScript: Handles real-time interaction with the backend.

Visualization

1. GAN Model Outputs: Compare outputs from 5 trained GAN models to evaluate performance.
2. User Interface Preview: Showcase the dashboard design for user interaction.

---

Screenshots

1. GAN Model Outputs

Below are the outputs of 5 GAN models trained on the sketch-to-image dataset:

2. User Interface Preview

A glimpse of the interactive dashboard for drawing and viewing results:

---

Contributing

If you'd like to contribute to this project, feel free to:

• Open an issue.
• Create a pull request.

---

License

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

---

Demo

Experience the transformation from sketch to color in action!