V-Share Amirali Charania, Bharat Ravisekar {amiralic, rbharat}@cc.gatech.edu Abstract In the present day world, people are increasingly using their camera-enabled mobile phones to take photographs of places they visit. However, it is difficult to organize, navigate and share such huge collection of photographs without tagging proper context with them. We feel that if we associate spatial, temporal and user based contextual tags with such photographs, then they can be managed and shared better. We further foresee a number of exciting applications like Interactive Albums and Collaborative Tourism using such a system. The focus of this project is on implementing a simple prototype system to demonstrate the viability of the idea. This document is to serve as the project proposal. Motivation and Objectives With the popularization of digital photography and the availability of affordable mobile devices having in-built cameras, people are now capturing and storing far more photographs than ever before. These large collections of photographs are inherently difficult to organize, navigate and share. The problem seems to arise from the intrinsic distributed-ness of these large collections and the lack of proper photographic context of these photographs. Yet, viewing and sharing such collections has a broad social and practical importance. The problem seems even more trying if we consider that the context of the photographs are in fact readily available in the form of GPS location, system time etc. In this project, we are planning to combine spatial, temporal and manually entered text tags with photographs to automate the tasks of organization, management and sharing of photographs among social networks. Building on this basic idea, we foresee the following interesting applications: 1. Interactive Album: Typically, when we download photographs from mobile devices to our PC, we store them as files in a folder without any context. Instead, if we place the photographs on a map, then it would add to the experience of viewing the collection. We could even share such maps among friends and social networks. 2. Collaborative Tourism: While planning a trip to an unknown place, we typically refer standard tourism guides and websites. However, these are often the places of general interest, which might not be relevant to the interests of a particular user. Suppose we have a social network of users with similar interests who share their travel experiences via geo-tagged photos. Then, we could access to information on “interesting” places like their location, photographs, detailed account of personal experience, most conducive times to visit etc from the system and be able to make a customized plan based on places that interest us. Objective: In this project, we plan to build a prototype of our idea and if possible deploy it on a java-enabled mobile phone that has a camera phone and GPS receiver. Related work Geo-tagging has been used before for applications that organize images for better exploration. Kentaro Toyama et. al. have used geo-tagging images for creating an indexed database of images and built an browsable application on top of it for easy navigation. Alexandar Jaffe et. al. have devised an algorithm to summarize and rank photographs based on their geo-tagging and other features to select best representative photographs. Our system would be an end-to-end application that spans the whole cycle: right from the moment a photograph is captured via a camera, to automatically tagging it and synchronizing it with the central repository to planning a trip. Besides, we plan to group users based on their social networks and interest groups for sharing pictures. Proposed work Steps: 1. Tagging of photographs: a. Location tag: The location tag (GPS coordinates) will be obtained using location-based APIs (JSR179) on a J2ME phone with a GPS receiver. We would append this location with the image header. An alternate option we might consider is to use a .NET CF instead of J2ME phones. b. Temporal tag: The temporal tag (date and time) will be obtained from the system as well as from the GPS satellites. We would append this with the image header. c. Manual tag: The manually entered text tag would be used by users to input their personal experiences and the semantic location information about the photograph. We will add this with the image header. 2. Storing of photographs: The photographs to be shared will be stored on a centralized database system exposed through a web-service. This would be done using a mobile client application which would communicate with the server side web services (.NET web services). The authentication of users would be based on a mapping stored on the server that maps the user to the groups to which he belongs. 3. Browsing of photographs: The photographs stored on the centralized database system will be browsed using maps as an interface. This would be a client side desktop application that uses map-point APIs to fetch the maps and web services to access the centralized database of photographs. The authentication of users would be based on a mapping stored on the server that maps the user to the groups to which he belongs. 4. Sharing the photographs: The user can download tagged photographs taken by other members of his group onto a mobile phone for on-tour reference. This would be a mobile client service using the .NET web services. The authentication of users would be based on a mapping stored on the server that maps the user to the groups to which he belongs. System Architecture: Plan of Action We intend to use the following technologies: 1. J2ME emulator WTK 2.5 - We would use J2ME emulator WTK 2.5 for developing our mobile client applications. 2. .NET – We would use Visual Studio .NET 2005 for developing web services and desktop client application in C#. 3. XML – We would use XML based data storage at the central image repository. 4. Map point– We intend to use Microsoft MapPoint web service for fetching the maps. Time Line The timeline that we intend to follow for this project is: Task a. b. c. d. e. Detailed Design Implementing the mobile client application on the emulator Implementing the server side application Implementing the desktop client application for browsing Evaluation and testing weeks 2 2 2 2 2 Evaluation and Testing Method Our system will require two client side applications and 1 server side application. We do not intend to implement the voice to text converter module for voice tags, Bluetooth synchronization module for synchronization, both of these would need far more work. Due to lack of hardware at this moment, we intend to demonstrate our system on an emulator. The evaluation should be based on how well we would have managed to achieve: 1. Tagging of photos with the spatial, temporal and user-entered tags. 2. Synchronization of the photos with the server. 3. Managing and organizing photos 4. Browsing of photos using maps 5. Downloading and viewing of the information for later reference while on tour Future Work 1. Tagging of photographs: Voice can be used as audio tags. These can replace manual tags which can be pain for the user on a mobile device. 2. Sharing of photographs: As per current system the sharing is done using shared centralized server over GPRS network but this can also be implemented using Bluetooth technology. The Bluetooth technology has a limitation of distance but advantage of peer to peer sharing. We foresee our application to be used on Parlay OSA framework and SIP protocol where it would be possible for peer to peer sharing of photographs over mobile phones using dynamic IP’s. Thus there would be no limitation of distance and advantage of peer to peer sharing as well. The Parlay OSA framework is already implemented in parts of Europe and ASIA. Bibliography 1. Alexander Jaffe, Mor Naaman, Tamir Tassa and Marc Davis - Generating summaries for large collections of geo-referenced photographs, WWW '06: Proceedings of the 15th international conference on World Wide Web year = 2006, isbn = 1-59593-323-9,pp 853-854, ACM Press 2. Kentaro Toyama, Ron Logan, Asta Roseway - Geographic Location Tags on Digital Images, Microsoft Research