Let's imagine that you have trained your model, integrated it with your app, and published it to the App Store. You can consider the job as done. But with neural networks, where there cannot be a situation when you will have a 100% accuracy superlightweight model, the job is never done.
You will continuously improve your dataset and retrain the model to improve its performance. And you will do it pretty often. Then you will need to put your model into your project and re-upload it to the App Store every time. Not very convenient.
In this video tutorial, our colleagues from skafos.ai show how you can train a model that detects a breed of a dog and then deploy this model directly to your app, without need to download the model and insert it to your Xcode project.