Wikitude

We want the least amount of steps users need to take in order to scan for targets.
This is achievable if the target is image-based because Wikitude have cloud recognition.
(By the least amount of steps, I mean, open ar scene and they can start scanning right away.)
With the current object tracking, I don't know how to make this possible. 

I understand the technical hurdle of having more than 10+ objects loaded into system memory for tracking. But is there any solution that can scale with object tracking? (Like cloud recognition that can handle a large database of 1000+ image targets)
The problem we're facing is that we need to have more than 10+ objects but WTO can only have a maximum of 10 targets. WTO also has a filesize limitation of 20 MB. Also, switching WTO by pressing buttons or choose from a list is acceptable but we prefer not doing it since it contradicts the UX we have for our app. So I would like to consult for the best alternative first before considering taking those steps.