picassopaintedit

Neural style transfer on images using DeepAI API and Tensorflow Example

This application supports neural style transfer on images in two modes.

1. fasttransform - Using DeepAI API.
2. deeptransform - Using Neural Style Transfer with tf.keras

Major Dependency:

• DeepAI API

https://deepai.org/api-docs/#neural-style

• TensorFlow and Keras
• Celery

http://www.celeryproject.org/

• AWS EC2 instance for running deeptransform and S3 for storing images.

app.py

• Flask application to support fasttransform and deeptransform endpoints.

Input:

• key - Neural style transform request key.
• style_url - URL of the style image used neural style transformation.
• content_url - URL of the content image used neural style transformation.

Output

• key - Neural style transform request key.
• tranformed_image_url - URL of the neural style transformed image.

flask_celery.py

• deeptransform is handled asynchronously using flask_celery.

deeptransformimpl.py

• Starting point for handling asynchronous task for flask_celery.

neuralstyle.py

• Implementation of deeptransform based on Neural Style Transfer with tf.keras.

util.py

• Implements util functions for storing neural style transformed images on AWS S3 and sending deeptransform notification.