There is provided a system and method for generation of vision recognition references from user-selected content. The method comprises receiving data corresponding to a user selected content, generating a vision recognition reference from the data, associating a game behavior descriptor with the vision recognition reference, wherein the game behavior descriptor correspond to a game behavior instance, and storing the vision recognition reference and the associated game behavior descriptor in a database, wherein the database includes a plurality of pre-defined vision recognition references, and wherein each of the plurality of pre-defined vision recognition references correspond to at least one pre-defined game behavior descriptor. The method may further comprise receiving user input corresponding to the user-selected content, retrieving the vision recognition reference and the associated game behavior descriptor from the database and initiating the game behavior in a corresponding game.