Summary
In this chapter, we introduced the concept of object detection, comparing it with object recognition and object localization. While the other two are limited to a single dominant object, object detection allows multi-object classification, including predicting their bounding boxes. We then spent some time introducing one particular algorithm, YOLO, before getting acquainted with Apple's Core ML Tools Python package, walking through converting a trained Keras model to Core ML. Once we had the model in hand, we moved on to implementing YOLO in Swift with the goal of creating an intelligent search application.
Despite this being a fairly lengthy chapter, I hope you found it valuable and gained deeper intuition into how deep neural networks learn and understand images and how they can be applied in novel ways to create new experiences. It's helpful to remind ourselves that using the same architecture, we can create devise new applications by simply swapping the data we train it on. For...
