Bio: Woontack Woo is an academic researcher from KAIST. The author has contributed to research in topics: Augmented reality & Context (language use). The author has an hindex of 32, co-authored 310 publications receiving 4070 citations. Previous affiliations of Woontack Woo include University of Southern California & Gwangju Institute of Science and Technology.
Papers published on a yearly basis
••15 Sep 2008
TL;DR: This work presents a method that is able to track several 3D objects simultaneously, robustly, and accurately in real-time in order to take the advantages of the two approaches to object detection and tracking.
Abstract: We present a method that is able to track several 3D objects simultaneously, robustly, and accurately in real-time. While many applications need to consider more than one object in practice, the existing methods for single object tracking do not scale well with the number of objects, and a proper way to deal with several objects is required. Our method combines object detection and tracking: Frame-to-frame tracking is less computationally demanding but is prone to fail, while detection is more robust but slower. We show how to combine them to take the advantages of the two approaches, and demonstrate our method on several real sequences.
••01 Jun 2012
TL;DR: This paper presents an itinerary model in terms of attributes extracted from user-generated GPS trajectories, and presents a social itinerary recommendation framework to find and rank itinerary candidates and compares mobile-only approach with Mobile+Cloud architecture for practical mobile recommender deployment.
Abstract: Planning travel to unfamiliar regions is a difficult task for novice travelers. The burden can be eased if the resident of the area offers to help. In this paper, we propose a social itinerary recommendation by learning from multiple user-generated digital trails, such as GPS trajectories of residents and travel experts. In order to recommend satisfying itinerary to users, we present an itinerary model in terms of attributes extracted from user-generated GPS trajectories. On top of this itinerary model, we present a social itinerary recommendation framework to find and rank itinerary candidates. We evaluated the efficiency of our recommendation method against baseline algorithms with a large set of user-generated GPS trajectories collected from Beijing, China. First, systematically generated user queries are used to compare the recommendation performance in the algorithmic level. Second, a user study involving current residents of Beijing is conducted to compare user perception and satisfaction on the recommended itinerary. Third, we compare mobile-only approach with Mobile+Cloud architecture for practical mobile recommender deployment. Lastly, we discuss personalization and adaptation factors in social itinerary recommendation throughout the paper.
••26 Oct 2010
TL;DR: This paper proposes a smart recommendation for highly efficient and balanced itineraries based on multiple user-generated GPS trajectories that users only need to provide a minimal query composed of a start point, an end point and travel duration to receive an itinerary recommendation.
Abstract: Traveling to unfamiliar regions require a significant effort from novice travelers to plan where to go within a limited duration. In this paper, we propose a smart recommendation for highly efficient and balanced itineraries based on multiple user-generated GPS trajectories. Users only need to provide a minimal query composed of a start point, an end point and travel duration to receive an itinerary recommendation. To differentiate good itinerary candidates from less fulfilling ones, we describe how we model and define itinerary in terms of several characteristics mined from user-generated GPS trajectories. Further, we evaluated the efficiency of our method based on 17,745 user-generated GPS trajectories contributed by 125 users in Beijing, China. Also we performed a user study where current residents of Beijing used our system to review and give ratings to itineraries generated by our algorithm and baseline algorithms for comparison.
TL;DR: The proposed model affirms the independence between sensor and application by using a unified context in the form of Who (user identity), What (object identity), Where (location), When (time), Why (user intention/emotion) and How (user gesture), called 5W1H.
Abstract: Context-aware application plays an important role in the ubiquitous computing (ubiComp) environment by providing the user with comprehensive services even without any explicitly triggered command. In this paper, we propose a unified context-aware application model which is an essential part to develop various applications in the ubiquitous computing environment. The proposed model affirms the independence between sensor and application by using a unified context in the form of Who (user identity), What (object identity), Where (location), When (time), Why (user intention/emotion) and How (user gesture), called 5W1H. It also ensures that the application exploits a relatively accurate context to trigger personalized services. To show usefulness of the proposed model, we apply it to the sensors and applications in the ubiHome, a test bed for ubiComp-enabled home applications. According to the experimental results, without loss of generality, we believe it can be extended to various context-aware applications in daily life.
21 Sep 2008
TL;DR: The set of accepted papers constitutes the largest number of papers accepted to a UbiComp conference to date, representing an overall acceptance rate of 19%.
Abstract: UbiComp 2008, the Tenth International Conference on Ubiquitous computing, is the premier forum for the presentation of research results in all areas relating to the design, implementation, deployment and evaluation of ubiquitous computing technologies. The conference brings together leading researchers, from a variety of disciplines, perspectives and geographical areas, who have been exploring the implications of computing as it moves beyond the desktop and becomes increasingly interwoven into the fabrics of our lives. A total of 226 submissions---160 full papers and 66 notes---were evaluated by our international Program Committee, composed of 31 of the leading researchers in the field of ubiquitous computing. Each submission was reviewed by two members of the program committee and two or more external reviewers, and after an online discussion phase 80 papers were chosen for consideration at a 2-day PC meeting. Each of these was evaluated by an additional PC member before the meeting. After this rigorous process, a total of 42 submissions---37 full papers and 5 notes---were accepted for publication in these proceedings, representing an overall acceptance rate of 19%. We are pleased to note that the set of accepted papers constitutes the largest number of papers accepted to a UbiComp conference to date.
••18 Jun 2018
TL;DR: Zhou et al. as mentioned in this paper propose VoxelNet, a generic 3D detection network that unifies feature extraction and bounding box prediction into a single stage, end-to-end trainable deep network.
Abstract: Accurate detection of objects in 3D point clouds is a central problem in many applications, such as autonomous navigation, housekeeping robots, and augmented/virtual reality. To interface a highly sparse LiDAR point cloud with a region proposal network (RPN), most existing efforts have focused on hand-crafted feature representations, for example, a bird's eye view projection. In this work, we remove the need of manual feature engineering for 3D point clouds and propose VoxelNet, a generic 3D detection network that unifies feature extraction and bounding box prediction into a single stage, end-to-end trainable deep network. Specifically, VoxelNet divides a point cloud into equally spaced 3D voxels and transforms a group of points within each voxel into a unified feature representation through the newly introduced voxel feature encoding (VFE) layer. In this way, the point cloud is encoded as a descriptive volumetric representation, which is then connected to a RPN to generate detections. Experiments on the KITTI car detection benchmark show that VoxelNet outperforms the state-of-the-art LiDAR based 3D detection methods by a large margin. Furthermore, our network learns an effective discriminative representation of objects with various geometries, leading to encouraging results in 3D detection of pedestrians and cyclists, based on only LiDAR.
01 Jan 2002
TL;DR: In this paper, the authors discuss the role of education as an avenue to liberate student learning capacity and, by doing so, to help teachers take charge of their lives as teachers.
Abstract: Dedication Preface Foreword PART I: FRAME OF REFERENCE We begin with the idea of giving students the tools that increase their capacity for learning. The primary role of education is to increase student capacity for personal growth, social growth, and academic learning. Models of Teaching is an avenue to liberate student learning capacity and, by doing so, to help teachers take charge of their lives as teachers. CHAPTER 1: BEGINNING THE INQUIRY Creating Communities of Expert Learners On the whole, students are in schools and classes within those schools. Both need to be developed into learning communities and provided with the models of learning that enable them to become expert learners. We study how to build those learning communities. CHAPTER 2: WHERE MODELS OF TEACHING COME FROM Multiple Ways of Constructing Knowledge The history of teacher researchers comes to us in the form of models of teaching that enable us to construct vital environments for our students. Models have come from the ages and from teacher-researchers who have invented new ways of teaching. Some of these are submitted to research and development and how teachers can learn to use them. Those are the models that are included in this book. CHAPTER 3: STUDYING THE SLOWLY-GROWING KNOWLEDGE BASE IN EDUCATION A Basic Guide Through the Rhetorical Thickets We draw on descriptive studies, experimental studies, and experience to give us a fine beginning to what will eventually become a research-based profession. Here we examine what we have learned about how to design good instruction and effective curriculums. And, we learn how to avoid some destructive practices. CHAPTER 4: MODELS OF TEACHING AND TEACHING STYLES Three Sides of Teaching--Styles, Models, and Diversity We are people and our personalities greatly affect the environments that our students experience. And, as we use various models of teaching our selves -- our natural styles -- color how those models work in the thousands of classrooms in our society. Moreover, those models and our styles affect the achievement of the diverse students in our classes and schools. PART II: THE INFORMATION-PROCESSING FAMILYOF MODELS How can we and our students best acquire information, organize it, and explain it? For thousands of years philosophers, educators, psychologists, and artists have developed ways to gather and process information. Here are several live ones. CHAPTER 5: LEARNING TO THINK INDUCTIVELY Forming Concepts by Collecting and Organizing Information Human beings are born to build concepts. The vast intake of information is sifted and organized and the conceptual structures that guide our lives are developed. The inductive model builds on and enhances the inborn capacity of our students. CHAPTER 6: ATTAINING CONCEPTS Sharpening Basic Thinking Skills Students can develop concepts. They also can learn concepts developed by others. Concept attainment teaches students how to learn and use concepts and develop and test hypotheses. CHAPTER 7: THE PICTURE-WORD INDUCTIVE MODEL Developing Literacy across the Curriculum Built on the language experience approach, the picture-word inductive model enables beginning readers to develop sight vocabularies, learn to inquire into the structure of words and sentences, write sentences and paragraphs, and, thus, to be powerful language learners. In Chapter 19 the outstanding results from primary curriculums and curriculums for older struggling readers are displayed. CHAPTER 8: SCIENTIFIC INQUIRY AND INQUIRY TRAINING The Art of Making Inferences From the time of Aristotle, we have had educators who taught science-in-the-making rather than teaching a few facts and hoping for the best. We introduce you to a model of teaching that is science on the hoof, so to speak. This model has had effects, among other things, on improving the capacity of students to learn. We concentrate on the Biological Sciences Study Group, where for 40 years science teachers have shared information and generated new ideas. And, Inquiry training is a "best yet" model for teaching basic inquiry skills. CHAPTER 9: MEMORIZATION Getting the Facts Straight Memorization has had something of a bad name, mostly because of deadly drills. Contemporary research and innovative teachers have created methods that not only improve our efficiency in memorization, but also make the process delightful. CHAPTER 10: SYNECTICS The Arts of Enhancing Creative Thought Creative thought has often been thought of as the province of a special few, and something that the rest of us cannot aspire to. Not so. Synectics brings to all students the development of metaphoric thinking -- the foundation of creative thought. The model continues to improve. CHAPTER 11: LEARNING FROM PRESENTATIONS Advance Organizers Learning from presentations has almost as bad a name as learning by memorization. Ausubel developed a system for creating lectures and other presentations that will increase learner activity and, subsequently, learning. PART III: THE SOCIAL FAMILY OF MODELS Working together might just enhance all of us. The social family expands what we can do together and generates the creation of democracy in our society in venues large and small. In addition, the creation of learning communities can enhance the learning of all students dramatically. CHAPTER 12: PARTNERS IN LEARNING From Dyads to Group Investigation Can two students who are paired in learning increase their learning? Can students organized into a democratic learning community apply scientific methods to their learning? You bet they can. Group Investigation can be used to redesign schools, increase personal, social, and academic learning among all students, and -- is very satisfying to teach. CHAPTER 13: THE STUDY OF VALUES Role Playing and Public Policy Education Values provide the center of our behavior, helping us get direction and understand other directions. Policy issues involve the understanding of values and the costs and benefits of selecting some solutions rather than others. In these models, values are central. Think for a moment about the issues that face our society right now -- research on cells, international peace, including our roles in Iraq and the rest of the Middle East, the battle against AIDS, poverty, and who controls the decisions about pregnancy and abortion. Not to mention just getting along together. PART IV: THE PERSONAL FAMILY OF MODELS The learner always does the learning. His or her personality is what interacts with the learning environment. How do we give the learner centrality when we are trying to get that same person to grow and respond to tasks we believe will enhance growth? CHAPTER 14: NONDIRECTIVE TEACHING The Learner at the Center How do we think about ourselves as learners? As people? How can we organize schooling so that the personalities and emotions of students are taken into account? Let us inquire into the person who is the center of the education process. CHAPTER 15: DEVELOPING POSITIVE SELF-CONCEPTS The Inner Person of Boys and Girls, Men and Women If you feel great about yourself, you are likely to become a better learner. But you begin where you are. Enhancing self concept is a likely avenue. The wonderful work by the SIMs group in Kansas (see Chapter 3) has demonstrated how much can be accomplished. PART V: THE BEHAVIORAL SYSTEMS FAMILY OF MODELS We are what we do. So how do we learn to practice more productive behaviors? Let's explore some of the possibilities. CHAPTER 16: LEARNING TO LEARN FROM MASTERY LEARNING Bit by bit, block by block, we climb our way up a ladder to mastery. CHAPTER 17: DIRECT INSTRUCTION Why beat around the bush when you can just deal with things directly? Let's go for it! However, finesse is required, and that is what this chapter is all about. CHAPTER 18: LEARNING FROM SIMULATIONS Training and Self-Training How much can we learn from quasi-realities? The answer is, a good deal. Simulations enable us to learn from virtual realities where we can experience environments and problems beyond our present environment. Presently, they range all the way to space travel, thanks to NASA and affiliated developers. PART VI: INDIVIDUAL DIFFERENCES, DIVERSITY, AND CURRICULUM The rich countryside of humanity makes up the population of our schools. The evidence suggests that diversity enhances the energy of schools and classrooms. However, some forms of teaching make it difficult for individual differences to flourish. We emphasize the curriculums and models of teaching that enable individual differences to thrive. CHAPTER 19: LEARNING STYLES AND MODELS OF TEACHING Making Discomfort Productive By definition, learning requires knowing, thinking, or doing things we couldn't do before the learning took place. Curriculums and teaching need to be shaped to take us where we haven't been. The trick is to develop an optimal mismatch in which we are pushed but the distance is manageable. CHAPTER 20: EQUITY Gender, Ethnicity, and Socioeconomic Background The task here is to enable differences to become an advantage. The best curriculums and models of teaching do just that. In other words, if differences are disadvantages, it is because of how we teach. CHAPTER 21: CREATING AND TESTING CURRICULUMS The Conditions of Learning Robert Gagne's framework for building curriculums is discussed and illustrated. This content is not simple, but it is powerful. CHAPTER 22: TWO WORDS ON THE FUTURE The Promise of Distance Learning and Using Models of Teaching to Ensure that No Child is Left Behind. Afterword APPENDIX PEER COACHING GUIDES Related Literature and References Index
01 Jan 1979
TL;DR: This special issue aims at gathering the recent advances in learning with shared information methods and their applications in computer vision and multimedia analysis and addressing interesting real-world computer Vision and multimedia applications.
Abstract: In the real world, a realistic setting for computer vision or multimedia recognition problems is that we have some classes containing lots of training data and many classes contain a small amount of training data. Therefore, how to use frequent classes to help learning rare classes for which it is harder to collect the training data is an open question. Learning with Shared Information is an emerging topic in machine learning, computer vision and multimedia analysis. There are different level of components that can be shared during concept modeling and machine learning stages, such as sharing generic object parts, sharing attributes, sharing transformations, sharing regularization parameters and sharing training examples, etc. Regarding the specific methods, multi-task learning, transfer learning and deep learning can be seen as using different strategies to share information. These learning with shared information methods are very effective in solving real-world large-scale problems. This special issue aims at gathering the recent advances in learning with shared information methods and their applications in computer vision and multimedia analysis. Both state-of-the-art works, as well as literature reviews, are welcome for submission. Papers addressing interesting real-world computer vision and multimedia applications are especially encouraged. Topics of interest include, but are not limited to: • Multi-task learning or transfer learning for large-scale computer vision and multimedia analysis • Deep learning for large-scale computer vision and multimedia analysis • Multi-modal approach for large-scale computer vision and multimedia analysis • Different sharing strategies, e.g., sharing generic object parts, sharing attributes, sharing transformations, sharing regularization parameters and sharing training examples, • Real-world computer vision and multimedia applications based on learning with shared information, e.g., event detection, object recognition, object detection, action recognition, human head pose estimation, object tracking, location-based services, semantic indexing. • New datasets and metrics to evaluate the benefit of the proposed sharing ability for the specific computer vision or multimedia problem. • Survey papers regarding the topic of learning with shared information. Authors who are unsure whether their planned submission is in scope may contact the guest editors prior to the submission deadline with an abstract, in order to receive feedback.
01 Dec 1988
TL;DR: In this paper, the spectral energy distribution of the reflected light from an object made of a specific real material is obtained and a procedure for accurately reproducing the color associated with the spectrum is discussed.
Abstract: This paper presents a new reflectance model for rendering computer synthesized images. The model accounts for the relative brightness of different materials and light sources in the same scene. It describes the directional distribution of the reflected light and a color shift that occurs as the reflectance changes with incidence angle. The paper presents a method for obtaining the spectral energy distribution of the light reflected from an object made of a specific real material and discusses a procedure for accurately reproducing the color associated with the spectral energy distribution. The model is applied to the simulation of a metal and a plastic.