This app uses deep learning to compose one image using the style of another image.
This app uses docker to create two microservices, the first is a Tensorflow/Serving container aim to handle a Tensorflow DNN model that makes the style transfer. And, the second microservice is a Flask application that handles the web request and displays the results.
Some screenshots of the web app:
-
Open your browser in localhost on port 5000 and you will see this page:
-
Select a source and a style image to process. The program will mimic the style image into the source image:
- The first time you compose up the services, it could take a few minutes since a DNN model is downloaded and a couple of python libraries are installed.
- The chace for the processed image can be found in ./python/static/temp/. In case you want to re-visid the processed images or erase the files.
- For info about the DNN model, please visit: https://www.tensorflow.org/tutorials/generative/style_transfer