ArKit is Apple's framework for Augmented Reality. It can do amazing stuffs but it couldn't be used to add layers to the existing word, for example informations inside a museum because the position of the virtual objects is given by the GPS coords of the phone, thus the precision in the position of the objects is in the order of meters.
But, as shown in the video, using machine learning is possible to get a more precise position, in the order of centemeters, even where the GPS has no signal.
The aim of this project is to make an aumented reality apps in which every user can add and share layers to the real word, everywhere and in the most easy way possible.
Based on CoreML-in-ARKit by hanleyweng
Model: Inception V3
Language: Swift 4.0
Written in: Xcode 9.0
Content Technology: SceneKit
Tested on iPhone 6s running iOS 11.2.6 (15D100)
You'll have to download "Inceptionv3.mlmodel" from Apple's Machine Learning page, and copy it into your XCode project. (As depicted in the following gif)
If you're having issues, double check that the model is part of a target (source: stackoverflow).
The Objects are the real objects seen by the camera The anchor is an object that you set as the reference point the reference system in every sesson of ARKit The nodes are arkit nodes composed by a point and a the name of an object You can change the anchor using the change anchor button but the nodes will be resetted
Right now the orientation precision is given by by the precision of the compass, the more still you are the better the precision but it is planned the automatic addiction of multiple anchors to improve it