Defining, training, and testing the model

Download both the training and testing datasets from http://benchmark.ini.rub.de/?section=gtsrb&subsection=dataset. Let's look at the steps to build the project after you have downloaded the dataset:

  1. Begin by transforming the images present in the dataset using histogram equalization. This is essential as each image in the dataset may have a different scale of illumination. You can see from the following two images how images of the same traffic sign have very different illumination. Histogram equalization helps to normalize these differences and makes the training data more consistent:

Once ...

Get TensorFlow Machine Learning Projects now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.