StyleArt is an open-source iOS library that turns photos into artwork on-device using Core ML: it ships a set of pre-trained neural style transfer models and applies them to any image with a single method call. Released by iLeaf Solutions on GitHub, it has around 230 stars; it is a 2018-era library built for iOS 11 and Swift 4.
Neural style transfer — re-rendering a photo in the manner of a famous painting — was popularised by apps like Prisma, but doing it well on a phone in 2018 meant running neural networks locally instead of sending images to a server. StyleArt packages that capability for iOS developers: drop the library folder into an Xcode project, call StyleArt.shared.process with an image and a style, and get the stylised result back asynchronously. Six styles are bundled, including Mosaic, The Scream, Muse, Udnie, Candy and Feathers, converted from the classic neural-style models into Core ML format so they run on the device GPU/NPU pipeline with no network and no per-image cost. The project includes an example app showing the full flow. It targets older toolchains, so modern projects would need some updating, but it remains a tidy reference for on-device Core ML image processing.
The bundled models are converted versions of well-known academic style transfer networks, and the library handles Core ML model loading and image sizing internally. Developers can add further converted models to expand the style catalogue following the same pattern as the six included examples, making it a compact case study in packaging research models for production iOS use.
StyleArt is free and open source, with no paid tier or usage fees.
iOS developers interested in on-device machine learning and photo effects, especially those studying how Core ML style transfer apps are put together.