Deep Learning Methods Overview

We use two deep learning approaches: one single instance of YOLOv8, and an ensemble method (bagging) of several YOLOv8 models. The process for the deep learning approaches differ quite significantly from the traditional methods, especially because YOLOv8 detects the type of the traffic sign together with its bounding box, so there is no need to separate the two tasks.

Both methods will be described in detail in the next pages.