Neural Style Transfer

This repository contains the implementation of a Neural Style Transfer model that applies the artistic style of one image to the content of another image using deep learning techniques.

Table of Contents

• Introduction
• Features
• Usage
• Project Structure
• License
• Acknowledgements

Introduction

Neural Style Transfer is a technique in deep learning where the style of one image is applied to the content of another image to create a new, stylized image. This project uses a pre-trained convolutional neural network (CNN) to achieve this transfer by minimizing the content and style loss functions.

Features

• Transfer artistic styles from a style image to a content image.
• Use pre-trained VGG19 network for feature extraction.
• Customize the level of style and content blending.
• Supports high-resolution images.

Arguments

• --content_image: Path to the content image.
• --style_image: Path to the style image.
• --output_image: Path to save the output image.
• --iterations: Number of iterations for optimization (default: 1000).

Acknowledgements –

This project uses the VGG19 model pre-trained on the ImageNet dataset.

• Inspired by the original Neural Style Transfer paper by Gatys et al.