ML-examples is a repository of machine learning programs written in Python 3 using Tensorflow and Keras. The intent of this repository is to provide some interesting ML applications which demonstrate a range of models and techniques.
As the name implies, this is a neural network that is trained to classify 27x27 pixel hand-written digits. This example can be run with either a basic neural network or a convolutional neural network, both trained on the MNIST hand-written digits dataset. You can test the models yourself by running
digit_ui.py - a tkinter canvas window which you can draw digits onto and you can (when the 'run model' button is pressed) use either model to classify whatever is on the canvas.
This is a tkinter application which you can train to play rock, paper, scissors against by making gestures to the camera. First, take at least 20 pictures of each gesture (keeping your hands within the box). Then compile the dataset and press 'build', this will train the model for 30 epochs. After that you can then play against the computer (there's nothing intelligent about this, it's just random) as many times as you want by making gestures and pressing 'play'.