Artificial intelligence allows not only technical breakthroughs, but researchers also figured out many exciting ways to leverage algorithms for creative application: generating visual arts, texts, and music.

We also didn't stay away from the trend, and together with Mlvch, we have developed and improved machine learning algorithms for photo and video artistic style transfer. These algorithms can "repaint" any given photo or video as it was drawn by your favourite artist: from Picasso to Van Gogh. We developed one of the first products in this field with both B2B and B2C cases.
Python / Matlab / Mathematical Modelling / Web Development / OpenCV / Tensorflow / PyTorch
People are always trying to look cool and original in the feed, that's why filters that make the photo better or masks with makeup or animal faces are booming last years. This trend doesn't seem to be stopped, only being deeply adopted by Facebook, Instagram, Snapchat and other big companies. For our client, we have developed an app that can repaint your photo or even the video in the style of any famous artist in the smallest details!
We have deeply researched and implemented several approaches to the image and video style transfer and adapted the algorithms for the visual and artistic feel of our client. The open-source available solutions have several bottlenecks and pitfalls that make it not so straightforward to apply for real-world use cases. For instance, the granularity and sizes of the "strokes" drawn by the AI-artist require some tricks to be controlled. Another issue is in video generation - it's easy to process it frame-by-frame. Still, in the end, the video has to be smooth, which made us solve an additional optimization problem. We adapted and improved known solutions according to the product point of view and t he final solution was based on deep neural networks and could run on mobile devices in real-time.
Our client had launched a competitive product being one of the first on the market that had 100K+ downloads on the mobile app markets and adapted the solution for successful B2B sales for marketing purposes.

Apart from the working product itself, the significant applied research knowledge base was created related to the artistic style transfer technology development, which allows building new related products very quickly and avoid possible pitfalls.

Ukraien, Kiev
Cupertino str. 132
+44 330 027 2146
International House,
64 Nile Str, London, N1 7SR,
United Kingdom
© 2020 Neurons Lab LTD. Registered in England and Wales
Company Number 12265479