In this chapter, we learned the building blocks of the YOLO object detection method and learned how it can detect an object so quickly and accurately compared to other object detection methods. We learned about different evolutions of YOLO—the original version of YOLO, YOLO v2, and YOLO v3—and their differences. We used YOLO to detect an object in an image and video file, such as traffic signs.
We learned how to debug YOLO v3 so that it can generate correct outputs without crashing. We understood how to use pretrained YOLO to make an inference and learned the detailed process for using our custom image to develop a new YOLO model and how to tune CNN parameters to generate correct results. This chapter also introduced you to RetinaNet and how it uses the concept of a feature pyramid to detect objects of different scales.
In the next chapter, we will be learning...