NeuraCanvas - AI Powered Art Generation App 🎨

An AI-powered art generation app that transforms your photos by blending their content with unique artistic styles to create stunning, one-of-a-kind images.

📖 Overview

Neural Style Transfer (NST) is a technique which combines two images (content image for the object of the image and the style image from which only the style is extracted) into a third target image.

Basically, in Neural Style Transfer we have two images- style and content. We need to copy the style from the style image and apply it to the content image. By, style I basically mean, the patterns, the brushstrokes, etc.

Note

Want to know more about the NST Paper? Check out this repo here: NeuStyle

Let's have a look to an example below: (The left-most image is the content, central image is the style and the right-most image is the output.)

You can check out the deployed app here: NeuraCanvas

🚀 Getting Started

Clone the repository and install the required dependencies using the following commands:

git clone https://github.com/aman-chhetri/NeuraCanvas.git

cd NeuraCanvas

pip install -r requirements.txt

streamlit run app.py

Then, your default browser (localhost) should open the following page:

Steps to run the app:

• Go to sidebar dropdown and choose option "Click here to run app!".
• After that upload the content and style images respecctively.
• Then after that click "START" button to start the model.
• After a while around 40-50 secs, you can see the generated images.

If the image looks unoptimized, try increasing iterations number (or tune the weights of the loss function).

👏 Acknowledgements

These are some of the resources I referred to while working on this project. You might want to check them out.

• The original paper on neural style transfer by Gatys et al .
• PyTorch's tutorial on NST.
• The original paper on VGG19.

I found these repos useful: (while implementing the model)

• fast_neural_style (PyTorch, feed-forward method)
• neural-style-tf (TensorFlow, optimization method)
• neural-style (TensorFlow, optimization method)

I found some of the content/style images from here:

• Rawpixel
• Wikimedia
• Unsplash
• Pexels

🤝 Feedback and Contribution

It is publicly open for any contribution. Bugfixes, new features, and extra modules are welcome.

• To contribute to code: Fork the repo, push your changes to your fork, and submit a pull request.
• To report a bug: If something does not work, please report it using GitHub Issues.

📬 Connect

• Email: chhetryaman3@gmail.com
• X: @amanxtwts
• Website: Portfolio
• LinkedIn: @amankshetri

© 2025 Aman Kshetri 👨‍💻