Put colors in your old photos with deep learning – new iCoService on board !

iCoSys extends its deep learning micro-services library with Colorizer, a service able to re-colour black and white photographies. Check it out at here.

Colorizer includes several interesting features. First, a slick and intuitive web showcase is there to allow you easily drag and drop black and white images and re-colour them in an instant. Second, the service exposes two deep network models: stable for realistic outputs and artistic for more fancy colouring. Third, we expose the service not only in the showcase, but also through a REST API (OpenAPI v3 OAS3), allowing you to script and use the service in your own applications.

Try it out ! Everything runs on kubernetes and is demonstrated on our platform. This platform is specifically tuned to support our research project and student education, including scalable ressources and gpu support. On a more technical note, the deep learning models are based on the pre-trained models available in the de-oldify project.

We congratulate Thomas Phung who built Colorizer in the context of his semester project while studying at the University of Applied Science in Fribourg. Houda Chabbi, Jean Hennebert and Jonathan Rial co-supervised the project. A special thanks to Jonathan for his help in the integration.

iCoSys is a research institute located in Fribourg, Switzerland and with a strong focus on applied Artificial Intelligence (AI).

Colorizer illustration. Einstein is now in colour.
Colorizer illustration. Einstein is now in colour.