This application uses a deep learning model trained on the famous MNIST dataset. Draw a number below, and watch the neural network predict it in real-time.
Trained on MNIST, a collection of 70,000 handwritten digits. Each image is a tiny 28x28 pixel grayscale grid.
Powered by a Sequential Dense Network. It flattens the 784 pixels into a 1D array and passes them through a 128-neuron hidden layer.
The final layer uses a Softmax activation function to output 10 probabilities, picking the digit (0-9) with the highest confidence.
Draw in the center of the box below, then click Predict.