Background: We would like to use visual recognition instead of a long menu system to identify different products. In this case different lego boxes. Selection of a lego box will lead to the appropriate build instructions for that toy and over assets becoming available within a lego app.

Assets: 5 kits, 6 images for each kit, one for each side of the packaging box.

Requirements:

(1) Use mobile phone and/or tablet camera/video to capture an image of the product packaging – this should work in any orientation and is intended to be captured by a child

(2) Compare this to an inventory of preloaded database assets and determine the product being captured.

(3) It should be possible to load additional assets into the database and to identify new packages without further configuration.

Figures below is from the Box 6 kit, contain 6 images from 6 different orientation