Create image classifiers on your iPad β no code required. Train models with your own photos, test them instantly, and export them for use in your apps or workflows. Optimized for mobile machine learning exploration.
Hi Product Hunt!
πI'm thrilled to introduce **MagicaLCore**, an iPadOS app that lets you train, test, and export image classifiers β all from your iPad, no coding required.
This project started from my own need to run fast, portable machine learning experiments. Whether you're a developer, student, or just curious about AI, you can:
π Import images or existing `.mlmodel` files
π§ Train your own classifier on device
π§ͺ Test it instantly with real-time predictions
π€ Export your model as a `.mlmodel` to use in your own appsBuilt natively in Swift for performance and ease of use.
I'd love to hear your feedback, suggestions, or just nerd out with you about machine learning on iOS! π
Thanks for checking it out π
β DaMy
Β Hi@rachitmagon, thanks so much for your interest!
Right now Iβm fully focused on improving and expanding the app on iOS. That said, Iβd love to bring it to Android in the future, as long as itβs possible to maintain the same level of stability, reliability, and smooth experience the app currently offers on iPad.
Supporting on-device model training is very dependent on hardware and OS-level APIs, and while iPad (especially with M1 or newer chips) makes this possible in a very efficient way, Iβll definitely explore the possibilities on Android if it becomes technically feasible.
Report
ππ Overflow AI launches today! Turn donation questions into real insights β smart giving just got smarter ππ§
Replies
TipTop Sheet
Smoopit
Hi @xdamyx .. this is pretty cool, are you planning to support android / google play as well to train models?
TipTop Sheet
Β Hi@rachitmagon, thanks so much for your interest!
Right now Iβm fully focused on improving and expanding the app on iOS. That said, Iβd love to bring it to Android in the future, as long as itβs possible to maintain the same level of stability, reliability, and smooth experience the app currently offers on iPad.
Supporting on-device model training is very dependent on hardware and OS-level APIs, and while iPad (especially with M1 or newer chips) makes this possible in a very efficient way, Iβll definitely explore the possibilities on Android if it becomes technically feasible.
ππ Overflow AI launches today! Turn donation questions into real insights β smart giving just got smarter ππ§