Book Image

Hands-On Vision and Behavior for Self-Driving Cars

By : Luca Venturi, Krishtof Korda
Book Image

Hands-On Vision and Behavior for Self-Driving Cars

By: Luca Venturi, Krishtof Korda

Overview of this book

The visual perception capabilities of a self-driving car are powered by computer vision. The work relating to self-driving cars can be broadly classified into three components - robotics, computer vision, and machine learning. This book provides existing computer vision engineers and developers with the unique opportunity to be associated with this booming field. You will learn about computer vision, deep learning, and depth perception applied to driverless cars. The book provides a structured and thorough introduction, as making a real self-driving car is a huge cross-functional effort. As you progress, you will cover relevant cases with working code, before going on to understand how to use OpenCV, TensorFlow and Keras to analyze video streaming from car cameras. Later, you will learn how to interpret and make the most of lidars (light detection and ranging) to identify obstacles and localize your position. You’ll even be able to tackle core challenges in self-driving cars such as finding lanes, detecting pedestrian and crossing lights, performing semantic segmentation, and writing a PID controller. By the end of this book, you’ll be equipped with the skills you need to write code for a self-driving car running in a driverless car simulator, and be able to tackle various challenges faced by autonomous car engineers.
Table of Contents (17 chapters)
1
Section 1: OpenCV and Sensors and Signals
5
Section 2: Improving How the Self-Driving Car Works with Deep Learning and Neural Networks
12
Section 3: Mapping and Controls

Summary

This has been a dense chapter, but hopefully you got a better overview of what neural networks are and how to train them.

We talked a lot about the dataset, including how to get correct datasets for training, validation, and testing. We described what a classifier is and we implemented data augmentation. Then we discussed the model and how to tune the convolutional layers, the MaxPooling layers, and the dense layers. We saw how training is done, what backpropagation is, discussed the role of randomness on the initialization of the weights, and we saw graphs of underfitting and overfitting networks. To understand how well our CNN is doing, we went as far as visualizing the activations. Then we discussed inference and retraining.

This means that you now have sufficient knowledge to choose or create a dataset and train a neural network from scratch, and you will be able to understand if a change in the model or in the dataset improves precision.

In Chapter 6, Improving...