In this repository you can find the code for the iOS app that we created in this post.
We use the BlazeFace architecture to build the model since this architecture is suitable for low-end hardware and mobile devices.
You can read more about the original model here.
The model is already in the repository due to its small size. You can also find a version of this app that uses TensorFlow Lite to run the model here.
This app supports multiple face detection and low-end devices like iphone and ipad devices with the A9 CPU.
The app uses SwiftUI to build the interface so there are a lot of code that could improve.