How to make an algorithm to detect unique objects

Question

I want to create an app for a museum that can recognize artifacts when a user holds up their phone camera to it (the camera would be built into the app). None of the APIs I found online seemed to have what I want. I'm hesitant to go straight into writing my own algorithm, but seeing as I don't really have an option, How would I write an algorithm to detect specific (unique) objects, given a database of pictures? Of course, if anyone has an easier solution than writing the algorithm myself, I would much appreciate that. I have not started anything on the app yet, I just need a general idea on how it would be possible so I can pitch the app idea.

Answer 1

I guess the best method would be to use deep learning to detect specific objects from your pictures. The path you can follow could be as follows,

• Collect a dataset of images and decide the artifacts to be detected
• Annotate your dataset i.e., label the images feasible for model training
• Train a deep learning model for object detection, on the prepared dataset

And then after testing the trained deep learning model, you can optimize or quantize the deep learning model ( according to your requirements) and integrate it into your app.