CS 8803-AIA Spring 2008 Project Proposal Krimy Amichandwala (krimya@gatech.edu) Nihar Gadkari (nihargadkari@gatech.edu) Samantha Misra (smisra6@gatech.edu) Urja Shah (urja.shah@gatech.edu) Introduction Napoleon Bonaparte once said that a picture is worth a thousand words. However, back in the 18th century, the French General definitely did not have the ease of access to high resolution scanners and digital cameras. With the increasing use and easy accessibility of high resolution cameras with no restrictions on memory, we introduce SnapWorks - a novel concept in sharing of a digital image library with your near and dear ones, or for that matter anyone who cares. SnapWorks introduces several features which will prove to be very useful to the average user who seeks to share his images with others. In addition to annotating images with comments, SnapWorks provides for several convenient features such as Clustering, Geotagging, Photo Bucket, and Photo Crunching. Motivation & Objectives The minutia of photo sharing is a tedious and complicated affair. We attempt to ease the trouble of sharing a vast photo collection providing several useful features along the way which the ordinary user would find strikingly refreshing. This document presents a brief outline of the application starting with the related work that inspired us and led to the conceiving of the concept and then moving onto the nittygritties of the system. Related Work There has been a lot of work in this area both in the industry and in academia. Few of the more well known commercial applications which help sharing of photos are Flickr.com, Google’s online photo sharing of Picasa web albums, Kodak Gallery, Yahoo Research’s ZoneTag etc. All of these have the same basic functionality which includes authentication of users, uploading albums of photos with some different features over and above the basic example, tagging of pictures, commenting, ordering prints of photos online, being location-aware etc. Also, there is ongoing research on geotagging and associating a picture with its location [1][3][4]. Considerable study has taken place in the ways of clustering photos in a meaningful way so as to provide users with new and improved ways of surfing their albums [2]. There are new devices which integrate a GPS system and a digital camera, e.g. the EverMore GT-800BT Bluetooth GPS EverPhoto[8]. Proposed Work We aim at building a system that allows users to upload photographs, share them with other users and collaboratively be able to manage and organize the photographs as per their needs. We plan to provide the following features in our system: 1. Sharing and Commenting: When a user shares his photos with another user, the other user can comment on the photos. These comments will be visible to the user who ‘owns’ the photo as well as other users who the owner has invited to view these photos. This will be a way for the users to interact and enjoy sharing of their photos. The feature of being able to comment on the photos can also be used in a professional setting where an architect needs to discuss his design of a building with an overseas client. A similar feature is implemented by the online social networking site, Facebook. Additional information about this can be found at [7]. 2. Clustering: This feature will enable the user to cluster his photos based on events, location and person. This will allow the user to create filters on his photos. If a user wants to know whether his friend was present for the Thanksgiving gathering in San Francisco, he will be able to do so by first filtering the photos by location. Doing so will only show him the photos clicked in San Francisco. He will then filter San Francisco photos by event. This will only show him the photos taken at the Thanksgiving gathering at San Francisco. Now he will be able to filter furthermore by specifying the friend’s name to the system who he wants to find. In order to provide the user with this facility we will incorporate a Face Recognition algorithm that will present the user with all the photographs within a set that contains a specified person. For this to be possible, the user will earlier provide the system with a sample of his friend’s photo in order for the system to know who it is searching for. We will either provide tagging facility (similar to Facebook’s tagging) for this or ask the user specifically for a photo of his friend. Clustering by location and event will be implemented as discussed in [3]. 3. Different views of photos: This feature will enable the user to view his photographs as thumbnails on a page (we refer to this as Bucket View) or thumbnails of photographs laid out on a map to provide the user with information of the photo density based on geographical locations. This feature is similar to geo-tagging implemented by Flickr. Below is a screen-shot from the same. Courtesy: www.flickr.com 4. Picture Panorama: We view this feature as an optimistic goal. Given a set of photos of a given location, the system will be able to generate a 360 o view of the location. This feature will provide a panoramic view of a place where a friend stays or even familiarizing a user of a location, he plans to visit. If two users have shared pictures of a monument, the system will stitch these photos together to give the user a panoramic view of the same. This idea has been implemented in Microsoft’s PhotoSynth [5]. Plan of Action Keeping in mind the project deadline, the timeline for the project will span 10 weeks, starting with this proposal as the first deliverable. The project schedule has been devised in terms of the five modules that we plan to implement. The completion of module in terms of milestones for the project is indicated below: Milestone 1: March 7th The basic application for ‘Photo-sharing’ will be completed and tested for various scenarios. The basic features will include the following: 1. Creation of albums 2. Sharing of photos by a user with contacts 3. Options to add comments by user or contacts Milestone 2: March 21st The enhancement to ‘Overlay photos on a Map’ will be completed. For this part of the project, we intend to enhance the existing ‘Carmen Sandiego’ system developed by a group of students in the Fall 2006 semester. In addition to viewing photos in the ‘Album’ format; the user will have the option of overlaying the picture on to a map based on the location of the picture. Milestone 3: April 4th The ‘Cluster Photos’ enhancement to the basic application will be completed and integrated as the next step in the project. This enhancement will be tested using photographs of the Georgia Tech campus. It will allow the user to cluster photos based on various parameters such as: 1. Locations 2. Tags 3. Persons Milestone 4: April 11th The ‘Photo bucket’ option will be completed and integrated with the application. Users will have the option of selecting from a number of different ways to view their photos. Milestone 5: April 25th The final enhancement of ‘photo crunching’ capability will be completed. The user will have the option to synthesize pictures to create 360o views. In addition, the application will be tested for any issues before the final submission. Evaluation and Testing Methods The application will primarily be useful for people who want to share photos or want to analyze the photos taking into consideration various factors like time, person or object involved in the photo or location where the picture was taken. Therefore while evaluating and testing the application, user friendliness and practical viability must be taken into consideration as a key feature. The components of the application which would be tested are: 1. 2. 3. 4. Photo sharing with comments Clustering Photo Bucket Picture crunching For photo sharing the users will be asked to login and invite a third party to view a particular album created by him. An uninvited user of the application will try to view this particular album in the testing phase. Only the invited users should be able to see the contents of the album. The user can also cluster the images in his/her respective albums according to location where the picture was taken, the time during which the picture was taken and the person or object who is in the picture. Various pictures which are correlated with respect to the parameters will be added into the different albums created by the user and the performance and effectiveness of the clustering algorithms will be evaluated. Also during the execution of some test cases seemingly disparate pictures will be added into the albums to check the effect on the clustering. We will also evaluate if any insight or knowledge is gained by the end user after forming these clusters. The photo bucket functionality will be tested by providing a virtual bucket (similar to a shopping basket) in which users can add multiple pictures. The users can then toggle between the bucket view and the map view where the pictures will be displayed on the specific point on the map. This functionality will be tested by adding large amount of photos into the basket and then toggling between views and to check whether these pictures are overlaid over the map appropriately. The users will not be given a detailed tutorial before using the application to test the usability of our application. This in part is because we will evaluate how a naïve user might cope with the application without having a background about photo sharing and image clustering. The feedback and the suggestions will be noted and we will try to make further improvements in the application if the need arises or if we feel it is applicable. In addition to user testing, testing methods to ensure that the application is robust for large files and large amounts of images will be implemented. Bibliography 1. “Geographic Location Tags on Digital Images”, Kentaro Toyama, Ron Logan, Asta Roseway, P. Anandan, ACM, 2003. 2. “Hybrid spatio-temporal structuring and browsing of an image collection acquired from a personal camera phone”, A. Pigeau, M. Gelgon, International Symposium on Image and Video communications over Fixed and Mobile Networks XXII: 5358, 2004. 3. “Automatic organization for digital photographs with geographic coordinates”, Mor Naaman, Yee Jiun Song, Andreas Paepcke, Hector GarciaMolina, ACM, 2004. 4. “Sharing, Discovering and Browsing Geotagged Pictures on the Web”, Carlo Torniai, Steve Battle, Steve Cayzer, hpl.hp.com, May 2007. 5. http://labs.live.com/photosynth/whatis/ 6. www.flickr.com 7. http://www.facebook.com/help.php?page=7 8. http://www.pocketgpsworld.com/evermore-geotagger.php