YOLO object detection models have been used in countless applications, from surveillance systems to autonomous vehicles. But what happens when you pair this capability of YOLOv8 with the KerasCV framework? Recently, KerasCV integrated the famous YOLOv8 detection model into its library. In this article, we will explore how to fine-tune YOLOv8 using a custom dataset. Along the way we will also cover the following points.
- Fine-tuning YOLOv8 on the traffic light detection dataset.
- Run inference on the validation image.
- Analyze the results.
Figure 1. KerasCV YOLOv8 output for traffic light detection.
Traffic light detection data set
We will train the KerasCV YOLOv8 model using the traffic light detection dataset. Thinklab's Small Traffic Light Dataset (S2TLD). The download link within the notebook provides a collection of images and annotations.
The dataset contains 4564 images with annotations presented in XML format. The following images clearly depict the different scenarios of collecting images.