IN ENGLISH: JavaScript mobile number figure-outer
ML: React Native implementation of the MNIST data trained in a BrainJS MLP (non conv) predictive model.
Due to a DOM dependency this project uses patch-package to modify Brain.js. This will automatically remove stream which is not polyfilled in React Native. The benefit? This project works out the box, no need to edit node_modules.
Additionally, there doesn't seem to be a way to get pixel data of images in React Native, so some native Objective C is included in this project. If requested, I'll create an Android version + node module of the getPixels function to complete cross-platform capabilities.
Special thanks to Robert Plummer and Ralph Schaer who's source helped speed this project along. Also, this would have been insanely difficult without the awesome node-module by Terry, the react-native-sketch-canvas.
Read more about MLP MNIST by Ralph here.
The MNIST dataset was published in 1999 by Lecun, read more here
As always, thanks to Infinite Red, for a supportive community while I work on adancing Machine Learning.