trained on 45 images of food and people with food. This is beta, as I'm sorting a dataset containing 200gb of images and therefore most likely will get the full version later