Current Projects
CAMAR 2.0 (Context-aware Mobile AR)
Recently, many mobile AR applications (e.g. AR browser, navigation application, and tour guide application) have been released. However, previous works focused on outdoor pseudo AR with sensory technology and visualizing 2D information related to recognized objects because of low visibility of a small screen. In addition to this, there have been increasing demands of the end-users to have more control in generating, sharing and consuming digital contents in various forms of virtual and augmented reality.
With this trend, CAMAR2.0 provides the intuitive and interactive way to make users to understand dynamic 3D information in situ with highly relevant to the target. Context-Aware Mobile AR 2.0 (CAMAR2.0) is next mobile AR technology which enables users to generate and to view social u-contents in seamless and personalized manner by utilizing integrated context information such as environmental changes as well as location, social relationship, and cultural differences of individual users.
To show the effectiveness of our work, we introduce an indoor AR tour guide application. To realize this, it recognizes region of planar objects with object recognition and tracking method in real time and manages users’ contextual information on a mobile platform. And then it augments dynamic personalized contents, and situational contexts and enables users to share the contents with other community members through their mobile devices.
MSRA-Project
Since the number of smart phone users is continuously increasing, there are also needs to browse social media on a mobile device for getting timely information on the site. However, it is still difficult to search and browse large amounts of data due to limitations of mobile devices which have small displays, low computation power, storage, and keyboards that are hard to use. So, it becomes important to effectively retrieve content related to the information needs of mobile users from web services.
Recently, several research projects related to “mobile + cloud” computing and context-aware computing have attempted to solve the difficulties. Because cloud computing can extends computation power and storage capacity and context-awareness can help understand mobile users’ information needs, combining the two technologies can improve mobile information retrieval by overcoming the limitations of mobile devices. When large amounts of data are used in information search and discovery and accessed by communities of users, a surprisingly successful access method is that of “tagging”, in which a user assigns a set of labels or tags to a document he is interested in. Tagging with contextual metadata is a good way to structure and access the large amount of data arising in cloud computing scenarios. Our project aims for ways to combine the strengths of hierarchical file organization, namely its optimal scalability, with that of tagging, which allows users fast and intuitive access to data.
To explore this research space we will develop a hierarchical context representation for tags designed to enhance the search and information discovery experience of mobile users by allowing them to do the social media retrieval based on current context such as location, time, and social network in mobile settings.
Infraware
Digilog book is a book which provides human’s five senses such as visual, haptic, auditory and so on. Since smart phone is catching popularity from public, Mobile Digilog book viewer is getting important. Generally, many computer vision algorithms are not performed on mobile devices. Therefore, The goal of this project is to make mobile digilog book tracker which supports many pages in real-time. Mainly, we are focusing on page recognition and tracking modules for mobile digilog book viewer. Furthermore, we will change the tracking target with 3D objects.
Global Frontier
Core technology for realtime augmentation of intelligent mirror world (1st level: Core technology for realtime convergence of interactive mirror world): To share immersive experience, connect the real and virtual world in realtime and detect/track multiple/large-scale physical objects (freeform and deformable objects), and then immersively augment in visual/auditory modality and provide intelligent reaction/behavior enabled 3D virtual objects in the dynamic real world, also the objects can be selected/manipulated by a user wearing a HMD.
