《Delivering-personalised-m-commerce-through-cloud-based-augmented-reality-on-billboards-View-Document》由会员分享,可在线阅读,更多相关《Delivering-personalised-m-commerce-through-cloud-based-augmented-reality-on-billboards-View-Document(2页珍藏版)》请在金锄头文库上搜索。
1、Delivering Personalised m-Commerce Through Cloud-Based Augmented Reality on Billboards Martn Lpez-Nores, Jos Juan Pazos-Arias, Yolanda Blanco-Fernndez, Manuel Ramos-Cabrer and Alberto Gil-Solla AtlantTIC Research Center for Information and Communication Technologies EE Telecomunicacin, Campus Univer
2、sitario s/n, 36310 Vigo (Spain) Email: mlnores,jose,yolanda,mramos,agildet.uvigo.es AbstractWe present a system that enables m-commerce interactions with billboards through augmented reality. A mo- bile application uploads image patterns to a recognition cloud, which asks a server to deliver mobile
3、applications to the users, customized according to his/her device and location, and per- sonalized according to his/her interests, preferences and needs. I. INTRODUCTION Augmented Reality (AR) entails a huge potential for adver- tising in various types of media, but it remains heavily under- exploit
4、ed because the AR-based advertising experiences have hitherto focused on the technical challenges of overlaying new contents on top of the live video coming from a camera 1. The interaction with the advertisers has been limited to linking media objects to URLs, so that a web browser pops up when the
5、 user taps on the screen. We present a system that enables much richer AR inter- actions with billboard advertisers, letting the users browse product catalogs (displayed on the billboards themselves), place orders, make appointments, etc. The interactions can be customized and personalized, meaning
6、that two billboards holding the same ad can be used to display different products and offers because of their different locations and the different profi les of the users viewing them. Likewise, the advertisers are given tools to monitor the interactions with their billboards, to update their catalo
7、gs, to choose the layouts and the look- and-feel of the interactive controls, etc. We describe the design of the system in Section II, followed by some implementation details and preliminary testing results in Sections III and IV. Conclusions are given in Section V. II. SYSTEM DESIGN Our system is b
8、uilt according to the principles of cloud com- puting, which makes it possible to relieve the clients from the burden of recognizing a particular billboard out of a potentially large set. Furthermore, the cloud provides convenient means to lodge the database of billboards, advertisers, catalogs and
9、users, along with the “Pattern recognition”, “Customization” and “Personalization” modules. A high-level diagram of the system is shown in Figure 1, with an example of a billboard featuring the drawing of a pizza boy. The “Lightweight client” (a mobile app) scans the picture and applies a simple fi
10、lter to create a mesh of relevant features, which is uploaded to the “Pattern recognition” module on the cloud. This module looks for a matching mesh in the ads database, and warns the “Customization” module to choose a convenient layout to render interactive controls and multimedia contents on the
11、users screen, taking into account the features of the device he/she is using. A pre-selection of products from the adver- tisers catalog can also be made at this point, considering the users location and the stock of the shops in the area. Then, the “Personalization” module may refi ne the selection
12、 according to the users profi le, which holds information related to his/her interests, preferences and needs. Once the customized and personalized service is running, the “m-Commerce” module keeps track of the users interactions to deliver notifi cations to the “Advertisers front-end”, a web applic
13、ation that allows the advertisers to manage orders, appointments, requests for information about some product, etc. Fig. 1. The main modules and fl ows of our system. III. IMPLEMENTATION DETAILS The “Lightweight client” has been built as an app for a mobile operating system using the APIs of an augm
14、ented reality framework, which natively provides the means to host 2015 IEEE International Conference on Consumer Electronics (ICCE) 978-1-4799-7543-3/15/$31.00 2015 IEEE162 the ads database and the “Pattern recognition” module on the cloud. The pattern matching algorithm can readily process the ful
15、l ad images, but it is possible to put a frame marker on one of the corners (the upper-left one in the example of Fig. 1) with encoded information about the advertiser itself, about the location of the billboard (as needed to provide location- based information) and even to support the recognition o
16、f the advertisement in case the image does not have a suffi ciently distinctive mesh of features. The “Customization” and “Personalization” modules have been developed as an evolution of our previous works in the area of cloud-based recommender systems for m-commerce (see 2). The user profi les are lists of properties, including (i) demographic features like age, gender or marital status, (ii) conditions bound to consumption patterns, such as topics of interest inferred from