Neural Style Transfer

This project is primarily a React-based mobile application with a backend Flask server that uses Artistic Style Transfer to stylize images using Convolutional Neural Networks. It has made use of Tensorflow for the MobileNetv2 model for the fast arbitrary image stylization and react-native for the development of the mobile application.

Block diagram

Requirements

• Tensorflow - v2.3
• Hub model - https://tfhub.dev/google/magenta/arbitrary-image-stylization-v1-256/2
• react-native-cli - v2.0.1
• flask - v1.1.1

To run the app:

• run the flask server
  • go to the root directory
  • execute flask run --host=0.0.0.0
• start the react-native application
  • go to the root directory
  • execute react-native run-android --no-jetifier

Sample screenshot

References

1. Justin Johnson, Alexandre Alahi, Li Fei-Fei “Perceptual Losses for Real-Time Style Transfer and Super-Resolution”
2. Leon A. Gatys, Alexander S. Ecker, Matthias Bethge “A Neural Algorithm of Artistic Style”
3. TFLite StyleTransfer using MobileNetV2