In this short tutorial we'll use Vision Framework to add object detection and classification capabilities to a bare-bones ARKit project. We'll use an open source Core ML model to detect a remote control, get its bounding box center, transform its 2D image coordinates to 3D and then create an anchor which can be used for placing objects in an AR scene.
https://medium.com/@Rozengain/using-vision-framework-object-detection-in-arkit-c0b5366f465d