When art critics question Pablo Picasso's cubism style, Picasso answers "Every child is an artist. The problem is how to remain an artist once we grow up." Those noble artists in the history not only have left us mind blowing masterpieces but also presented us the freedom of expressing themselves. When we are creating MONA, we hope to build a bridge between those great artists and every one in our normal life. Reliving those artists' style in people's every day photos, we want to tell our users that every moment in your life is a pretty perishable and you are the artist of your life.

What it does

Empowered by machine learning technology, MONA captures the style (color tone, lighting, brushstrokes, texture, etc.) of fine-art masterpieces and apply the style in your own photos.

You can upload your photo to MONA web app and pick up an artist's painting (like VanGogh's Starry Night) and the painting style will be added to your photo.

How we built it

Our inspiration came from an academic paper on Computer Vision and Pattern Recognition. Starting from training data, we build our own deep style service with Python on our cloud hosting including the restful API. Then we embed the service and build a mobile friendly website with Javascript and Python.

Challenges we ran into

  1. The image size that our GPU can process is very limited
  2. The cloud processing time is not stable because the hosting's bandwidth is not so high.

Accomplishments that we're proud of

We are very happy to see that we have succeeded in filling the gap between technology and product. Our beta testers really enjoy playing MONA.

What we learned

  1. Prioritize the tasks when time and resources are limited
  2. Seamlessly integrate between design and development

What's next for MONA

Our next step is to first optimize our GPU and current machine learning algorithm to make it more powerful. For product, we plan to build a community where users can create their own filters and share their creation with others.

Built With

+ 12 more
Share this project: