http://smartech.gatech.edu/handle/1853/5978 Original work by students of College of Computing Search the Channel search http://smartech.gatech.edu/simple-search http://smartech.gatech.edu/handle/1853/29784 Title: Visual place categorization <br/> <br/>Authors: Wu, Jianxin <br/> <br/>Abstract: Knowing the semantic category of a robot's current position not only facilitates the robot's navigation, but also greatly improves its ability to serve human needs and to interpret the scene. Visual Place Categorization (VPC) is addressed in this dissertation, which refers to the problem of predicting the semantic category of a place using visual information collected from an autonomous robot platform. Census Transform (CT) histogram and Histogram Intersection Kernel (HIK) based visual codebooks are proposed to represent an image. CT histogram encodes the stable spatial structure of an image that reflects the functionality of a location. It is suitable for categorizing places and has shown better performance than commonly used descriptors such as SIFT or Gist in the VPC task. HIK has been shown to work better than the Euclidean distance in classifying histograms. We extend it in an unsupervised manner to generate visual codebooks for the CT histogram descriptor. HIK codebooks help CT histogram to deal with the huge variations in VPC and improve system accuracy. A computational method is also proposed to generate HIK codebooks in an efficient way. The first significant VPC dataset in home environments is collected and is made publicly available, which is also used to evaluate the VPC system based on the proposed techniques. The VPC system achieves promising results for this challenging problem, especially for important categories such as bedroom, bathroom, and kitchen. The proposed techniques achieved higher accuracies than competing descriptors and visual codebook generation methods. http://smartech.gatech.edu/handle/1853/29771 Title: Supporting human interpretation and analysis of activity captured through overhead video <br/> <br/>Authors: Romero, Mario <br/> <br/>Abstract: Many disciplines spend considerable resources studying behavior. Tools range from pen-and-paper observation to biometric sensing. A tool's appropriateness depends on the goal and justification of the study, the observable context and feature set of target behaviors, the observers' resources, and the subjects' tolerance to intrusiveness. We present two systems: Viz-A-Vis and Tableau Machine. Viz-A-Vis is an analytical tool appropriate for onsite, continuous, wide-coverage and long-term capture, and for objective, contextual, and detailed analysis of the physical actions of subjects who consent to overhead video observation. Tableau Machine is a creative artifact for the home. It is a long-lasting, continuous, interactive, and abstract Art installation that captures overhead video and visualizes activity to open opportunities for creative interpretation. We focus on overhead video observation because it affords a near one-to-one correspondence between pixels and floor plan locations, naturally framing the activity in its spatial context. Viz-A-Vis is an information visualization interface that renders and manipulates computer vision abstractions. It visualizes the hidden structure of behavior in its spatiotemporal context. We demonstrate the practicality of this approach through two user studies. In the first user study, we show an important search performance boost when compared against standard video playback and against the video cube. Furthermore, we determine a unanimous user choice for overviewing and searching with Viz-A-Vis. In the second study, a domain expert evaluation, we validate a number of real discoveries of insightful environmental behavior patterns by a group of senior architects using Viz-A-Vis. Furthermore, we determine clear influences of Viz-A-Vis over the resulting architectural designs in the study. Tableau Machine is a sensing, interpreting, and painting artificial intelligence. It is an Art installation with a model of perception and personality that continuously and enduringly engages its co-occupants in the home, creating an aura of presence. It perceives the environment through overhead cameras, interprets its perceptions with computational models of behavior, maps its interpretations to generative abstract visual compositions, and renders its compositions through paintings. We validate the goal of opening a space for creative interpretation through a study that included three long-term deployments in real family homes. http://smartech.gatech.edu/handle/1853/29767 Title: Learning in public: information literacy and participatory media <br/> <br/>Authors: Forte, Andrea <br/> <br/>Abstract: This research examines new systems of information production that are made possible by participatory media. Such systems bring about two critical information literacy needs for the general public: to understand new systems in order to assess their products and to become adept participants in the construction of public information spaces. In this dissertation, I address both of these needs and propose a view of information literacy that situates the information literate as both consumer and producer. First, I examine a popular example of a new publishing system, Wikipedia, and present research that explains how the site is organized and maintained. I then turn my attention to the classroom and describe three iterations of design-based research in which I built new wiki tools to support publication activities and information literacy learning in formal educational contexts. I use the rhetorical notion of genre as an analytic lens for studying the use and impact of these new media in schools. Classroom findings suggest that the affordances of a wiki as an open, transparent publishing medium can support groups of writers in building a shared understanding of genre as they struggle with an unfamiliar rhetorical situation. I also demonstrate how writing on a public wiki for a broad audience was a particularly useful writing experience that brought about opportunities for reflection and learning. These opportunities include transforming the value of citation, creating a need to engage deeply with content, and providing both a need and a foundation for assessing information resources. http://smartech.gatech.edu/handle/1853/29750 Title: Understanding the social navigation user experience <br/> <br/>Authors: Goecks, Jeremy <br/> <br/>Abstract: A social navigation system collects data from its users--its community--about what they are doing, their opinions, and their decisions, aggregates this data, and provides the aggregated data--community data--back to individuals so that they can use it to guide behavior and decisions. In this thesis, I document my investigation of the user experience for social navigation systems that employ activity data. I make three contributions in this thesis. First, I synthesize social navigation systems research with research in social influence, advice-taking, and informational cascades to construct hypotheses about the social navigation user experience. These hypotheses posit that community data from a social navigation system exerts informational influence on users, that users egocentrically discount community data, that herding in social navigation systems can be characterized as informational cascades, and that the size and unanimity of the community data correspond to the strength of the community data's influence. The second contribution of this thesis is an experiment that evaluates the hypotheses about the social navigation user experience; this experiment investigated how a social navigation system can support online charitable giving decisions. The experiment's results support the majority of the hypotheses about the social navigation user experience and provide mixed evidence for the other hypotheses. The implications that arise from the experiment's findings compromise the final contribution of this thesis. These implications concern improving the design of social navigation systems and developing a general framework for evaluating the social influence of social navigation systems.