Generative Adversarial Networks to Improve the Nature of Training in Autonomous Vehicles

In this paper, the Generative Adversarial Network is utilized to generate more training samples from the real-time environment. This increases the nature of training the model used for automated prediction in deep learning model for autonomous vehicle applications. The generated model enables better classification accuracy and increases the robustness of the model. The training is improved in a manner for increasing the effectiveness in classifying the scenes of an autonomous intelligent system in modern cars. The deep vision is set up in python and instances generated by the Generative Adversarial Network offers constrained generation of samples for optimal classification of scenes. The results show that the intelligent system offers a better classification of the scenic environment than the conventional deep vision models.

I. Introduction

Deep learning and autonomous vehicles are two of the most rapidly advancing fields of technology. Autonomous vehicles rely on the accuracy of deep learning algorithms to accurately classify the environment around them. Deep learning is one of these areas. Deep learning algorithms are an absolute requirement for the successful development of autonomous cars. This is because these algorithms can accurately categorize the settings in which they operate [1].