A contextual local image recognition module of a device retrieves a primary content dataset from a server and then generates and updates a contextual content dataset based on an image captured with the device. The device stores the primary content dataset and the contextual content dataset. The primary content dataset comprises a first set of images and corresponding virtual object models. The contextual content dataset comprises a second set of images and corresponding virtual object models retrieved from the server.