so I have this dataset of images of people sitting in a restaurant.
I've annotated about 300 images with an average of 30 instances of "person" per image.
Now I'm wondering if I should have annotated only one (or just a few) person per image and processed way more images ?
I've successfully trained an SSD network with only one class, but I'm still wondering if I should have gone the other way...
Anyone got input on that ?