Detecting custom objects in images, and online API -- 2

I will provide pictures different monuments, statues, buildings, etc. And your job will be to develop a model for object location, and put online an API for sending requests. Workflow would be like this: -Picture is sent to API -API answers wether a known object has been detected, which one and how confident it is. It's important to have a possibility where you send a picture and it tells you it's not in the database. Therefore sending the best match among the known monuments is not an option. I'll appreciate specific interest and not just a generic answer. The API should be online during the summer, after that we would talk again. Database for the pilot won't be bigger than 20 elements (with around 300 pics each).