ABSTRACT
The first autonomous cars developed by Carnegie Mellon University and ALV appeared in 1984. There are many datasets built specifically for testing autonomous driving (AD) applications. They are divided into two main categories: real and synthetic. In addition to geometric information, there is a very important information prior that can be used in AD applications. The chapter focuses on computer vision tasks for AD applications. It explores the AD field and the role of recent computer vision algorithms in self-driving car development, such as deep learning convolutional neural network (CNN), which showed significant improvement in terms of accuracy for various tasks in different applications. CNNs and in particular deep learning have played a great role in computer vision, and they are exploited in various applications for AD. The chapter presents a brief introduction about deep CNN, i.e., how neural networks are used in machine learning, and the idea behind it.
