Need some validation on my idea

So I have been doing web dev work for a while and recently I started learning about machine learning and AI because who doesn't want to build their own Jarvis am I right. This idea just came to me the other day and after thinking through how I would build it, I believe I can create it given a few days. So here's my idea.

Custom image recognition software as a service. Let me dive a bit deeper into it. Image recognition is up and coming and is out there for sure, but to get it to work on very specific pictures, it'll have to be fine tuned. I was thinking of creating a service where you can build your own fine tuned image recognition model to deploy however you want in your business. All you would need to provide is a good set of labeled specific pictures for what you want and my app would handle the rest.

Here's an example. Say for whatever reason your car business needs a software that can detect with decent certainty what kind of Tesla picture your user just uploaded. So, you'd sign up on the website, provide a data set of different pictures of Model X, Model Y, etc. My app would build a specific model to detect just that - The types of Teslas, and bam it's ready to use. Then I'd have an API where all you have to do is upload the picture and I'd output what kind of Tesla it is.

I'm not sure if this is a good idea or if I can get it to work with amazing accuracy, but I think I can do it. In any case, would this be interesting or something you think people may use?

PS. If you have any alternate idea along this line or a completely different idea that uses machine learning (and is not insanely difficult) you'd like to see, please let me know! Not trying to build the next robot but I'm just trying to find practical project ideas to build and learn with. Thanks!