I have retrained ssd_mobilenet_v2 using my own dataset with 2 classes (pen or pencil), using object detection API. For my project, I expect users to select specific pencils from all pencils and further retrain the network according to user-chosen pencils.
My question is how big should be a dataset of new pencils be? And should the user choose an object that is not desired (pens in this case)?
The same solution should cover if users want to choose a specific type of pen.