Abstract
Autonomous driving has received much attention in academia and industries in recent years. Researchers incorporate different perceptions to detect vehicles, objects, routes, and lighting. However, few publications focus on road signs on pavement surface. This paper aims at creating a dataset with 10,000 images of U.S. road markings from similar datasets, web scraping, and raw driving footage in California. In addition, this paper proposes an ensemble deep learning model as a component to autonomous driving systems to allow real time detection of road markings and other objects. The proposed ensemble model is based on four modified models, including improved CenterNet, improved EfficientDet-D1, improved SSD and improved YOLOv4. Finally, the report experiments show that the proposed ensemble model classification of mobile objects on roads for autonomous vehicles.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have