On the Applicability of Deep Learning for Road Signal Recognition

Deep Learning is part of a broader family of machine learning methods based on learning data representations that is being widely deployed in autonomous vehicles with the purpose to identify and recognize objects on the environment in surroundings. There are several methods and strategies which allow the workflow of training and validation to a specific case to recognize objects. Moreover, how is the right way to apply deep learning in a case of for instance to recognize signals, pedestrian, cars and so on? We show a whole workflow that demonstrate how to make the correct deployment of the deep learning since the dataset definition, training, validation, calibration and performance analysis. This way, we hope to provide some important aspects that should be relevant in the deep learning process. An example of the traffic signal recognition is conducted and the results present effectivness and satisfactory performance for the trained network.