At first it was a rough implementation to see if the tech stack could do what I want; later it was "let's polish to make it usable by others"; later it was "I need to make it good enough that paying customers would be very pleased". Numerous times I kept wanting to just call it finished, but there were always cool features to add.
I've learned a great deal, and it's immensely satisfying to work on. I've explored face recognition and server capabilities (learning a lot and having a good time). And eventually will likely add both features to the released product.
I'm sure there is great value in small exploratory projects. But I really hope that people polish things up a bit more (at least a screenshot on the GitHub page!) and share it with the world in a more-finished form. There's value in that too.
Cheers!