I'm in the middle of trying to get Tensorflow Lite project working and even though I'm a complete beginner at Azure Custom Vision Service, I was following a tutorial and went into researching more links that I'll display soon. FIRST, let me show you what the azure vision service new project interface looks like at the moment so you understand the problem.
Now for the first tutorial I was reading the link is here. A very important sentence in this article states "Make sure you select one of the compact domains, the compact domains create models that are small enough to be exported and used from a mobile device.". Clearly I don't see any "domain" options available from the pictures I've shown in my own attempt at the project set up.
I found other links that asserted getting a compact domain was absolutely necessary like this YouTube video for example at 11:40. He explains "regular" vs "compactable" but says to make sure to select compact or it won't work.
I have a few questions/concerns.
Why doesn't my project creation interface blatantly give me any domains, such as compact, to select from? I have to be sure of this according to links I've looked into.
In "Create New Resource" there's a "Kind" section with "CognitiveServices", "CustomVision.Training", and "CustomVision.Prediction". I also went across a different article/video that I can't, for the life of me, find at this moment but it did state the importance of being able to train and predict. So this makes me scratch my head as well. What would I select here as a beginner that's looking to of course train a model and run predictions? Is that what the option "CognitiveServices" does?
Always grateful for anyone to take their time to help. It would be much appreciated. Thank you.
So this got resolved by me actually creating the custom vision service in portal.azure.com and THEN going back to customvision.ai (logging in with the same account of course).
Problem solved. Hopefully this benefits some people in the future, none of the resources I found looking for help made it clear that I have to create the custom vision service first.