Hi everyone
This is my first post here and I hope that the subforum is the right one. I wanted to retrain an image classifier to recognize my keys, i.e. physical keys as you use them to open and lock doors in a building. I have never done this, but according to some articles I have to retrain the latest layer of the pretrained model. And for this I need a huge dataset with objects similar to the ones that I want to classify. Therefore I wanted to use a program that automatically downloads images of keys from Google. So far so good. But as I understood it, these keys have to be labelled, but I do not know in which way. I do not know these keys. And in the end I just want to photoscan my own keys, tell app which is which and when I later scan a key again, it should recognize it and tell me which key it is.
Am I on the right track?
Best regards
Martin