Digit recognizer in C#/Javascript, with a optional Pickle REST service and an ONNX Mnist Machine learning model for predictions created in Python.
- Uses Python code in a Jupyter notebook to generate an Mnist machine learning model file in ONNX and Pickle format, with training data from the Mnist Dataset and 5 different algorithms (Logistic Regression, K-Nearest Neighbors, Random Forest, Extremely Random Trees, and a Convolutional Neural Network).
- Uses C# code to load that ONNX model into an ML.NET ONNX InferenceSession variable.
- Uses Javascript and HTML to allow users to draw a number in an HTML canvas (drawingboard.js), and have an AJAX call activate the model and try to recognize the correct answer through either a Controller Action with ONNX, or a REST Service with Pickle.
- Uses Python code with FLASK