Image Quantization using CUDA-Accelerated K-Means Clustering

Overview

This project implements K-Means Clustering for image quantization using CUDA acceleration to significantly improve performance. The application is built with PyTorch (CUDA), NumPy, OpenCV, and Tkinter for an interactive GUI.

Features

• GPU-Accelerated K-Means Clustering using PyTorch CUDA tensors.
• Tkinter GUI for easy image selection and quantization.
• Adjustable Number of Colors to control the level of quantization.
• Performance Timer to display execution time.
• Supports Various Image Formats including JPG, PNG, and BMP.

Technologies Used

• Programming Language: Python 3
• Libraries: PyTorch, NumPy, OpenCV, Tkinter, PIL (Pillow)
• CUDA Support: GPU acceleration for fast clustering

Installation

Prerequisites

Ensure you have the following installed:

• Python 3.10+
• CUDA-compatible GPU
• PyTorch with CUDA
• Required Python libraries

Installation Steps

1. Clone the repository:

   git clone https://github.com/The-Silly-Glitch/Color-Quantization-Using-Cuda.git
   cd Color-Quantization-Using-Cuda

2. Create a virtual environment (optional but recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

Usage

Running the GUI

To start the application, run:

python gui.py

Using the Application

1. Click Browse Image and select an image.
2. Enter the desired number of colors (clusters).
3. Click Quantize Image to process the image.
4. View execution time and preview the quantized image.
5. Click Save Quantized Image to save the output.

Performance Improvement

Image Size	CPU Execution Time	CUDA Execution Time
512×512	~5s	~0.8s
1024×768	~12s	~1.5s
1920×1080	~30s	~3.5s

Folder Structure

├── src
│   ├── gui.py               # GUI application
│   ├── kmeans_clustering.py # CUDA-accelerated K-Means
│   ├── image_processor.py   # Image loading and processing
├── requirements.txt         # Required dependencies
├── README.md                # Project documentation

Future Improvements

• Implement Mini-Batch K-Means for further speedup.
• Add support for custom distance metrics.
• Improve color accuracy with alternative initialization techniques.