Digit recognition
The digit recognition MNIST dataset was developed by Yann LeCun, Corinna Cortes, and Christopher Burges for assessing machine learning models on the handwritten digit problem. Digit images were taken from a mixture of scanned documents, normalized in size, and centered. Each image is 28 pixels in height and 28 pixels in width, for a total of 784 pixels in total. Each pixel has a single pixel value associated with it, indicating the lightness or darkness of that pixel, with higher numbers meaning darker. This pixel value is an integer between 0 and 255, inclusive. We develop a digit recognition pipeline. We have 10 digits (0 to 9), or 10 classes, to predict.
Getting ready
In this recipe, we develop a modeling pipeline that tries to recognize a digit (0-9) based on images with greater accuracy. The modeling pipelines use CNN models written using the Keras functional API for image classification.
The Keras library provides a simple method for loading the MNIST data. The dataset...