Recognition and classification based on the camera images of consumer products on the shelves. The project was realized using the Vispera Database which contains items loosely cropped from retail shelf images. The associated task consists of training a classifier over 180 classes to recognize new instances from these categories with a confidence mechanism, which can be used to rank potential class matches.