Object Detection for Autonomous Vehicles
Overview and Motivation
Autonomous vehicles have the potential to benefit society in multiple ways, including safety, convenience, and mobility [1]. In terms of safety, autonomous vehicles would remove human error from driving, reducing traffic accidents. Autonomous vehicles also provide convenience, as time normally spent driving can be spent on other tasks while in the vehicle. Last, autonomous vehicles could provide better mobility for people with disabilities who may have difficulty operating or are unable to operate a standard vehicle. However, one challenge in developing autonomous vehicles is object detection, which is used to detect pedestrians, other vehicles, traffic lights, and traffic signs, to name a few. Detecting these types of objects is a critical part of making the vehicle avoid collisions with people or other vehicles, obey stop signs, and stop at red lights, for example. Thus, the focus of this project is training multiple object detection models for autonomous vehicles and comparing their performance. All the code I wrote for this project is on GitHub. Below is a video summarizing the project.