| HOANG DANH LIEM | 1461209: D, 中間発表 | インタラクティブメディア設計学 加藤 博一, 横矢 直和, Christian Sandor, 武富 貴史 |
|
title: Realistic Interaction in a Room-Sized Environment with a Visuo-Haptic Augmented Reality System abstract: Visuo-Haptic Augmented Reality (VHAR) system allows users to see and touch digital contents that is embedded in the real environments with the presence of sensors which obtain the information from the world and displays that augment the environment with computer-generated visual and haptic stimuli. However, the current VHAR systems lack the realistic interaction between the virtual objects and the real world in term of visual and haptic feedback. In this research, we investigate the technical problems and suggest the solution for VHAR system which supports real-time interaction of virtual contents and the real environment in room-sized environment. Our method is to synchronize the reconstructed models from 2 depth cameras and use them for both graphic and haptic rendering. The result allows users to use the haptic device with virtual tools attached to it to feel the real objects as it is. language of the presentation: English | ||
| KRICHENBAUER MAXIMILIAN MICHAEL | 1461211: D, 中間発表 | インタラクティブメディア設計学 加藤 博一, 横矢 直和, Christian Sandor, 武富 貴史 |
|
title: Towards Immersive 3D User Interfaces for 3D Design
abstract: Using AR and VR technology to create User Interfaces for 3D modeling, animation, and other design tasks was repeatedly investigated in previous decades, but none of the research resulted in a usable work environment that appealed to 3D designers for adoption. In our research we worked closely with 3D designers to discover their requirement and how immersive 3D User Interfaces might possibly benefit their work. We developed a prototype system and used it to perform several user studies. language of the presentation: English | ||
| TEJERO DE PABLOS ANTONIO | 1461215: D, 中間発表 | 視覚情報メディア 横矢 直和, 加藤 博一, 佐藤 智和, 中島 悠太 |
|
title: Human action recognition-based personal sports video summarization abstract: Automatic sports video summarization poses the challenge of acquiring semantics of the original video, and existing work leverages various knowledge in application domains, e.g., structure of games and editing conventions. We propose a personal sports video summarization for self-recorded RGB-D videos, which became available to the public due to the commodification of off-the-shelf RGB-D sensors. We focus on sports whose games consist of a succession of actions and, unlike previous research, we use human action recognition on the depth sequences in order to acquire higher level semantics of the video. Human action recognition itself poses several challenges, esp. when we do not have enough instances to train our classifier. We approach this challenge by introducing a flexible human action recognition method able to recognize actions even with a small number of training instances, and able to learn new instances in runtime. We evaluated our proposed methods with very positive results. language of the presentation: English | ||
| LUHANGA EDITH TALINA | 1461212: D, 中間発表 | ユビキタスコンピューティングシステム 安本 慶一, 金谷 重彦, 荒川 豊, 諏訪 博彦, 藤本 まなと |
|
title: Implementing and evaluating effectiveness of theory-based behavior change strategies in weight loss applications
abstract: Behavior change is the most effective way to lose weight in the long-term. Around 20 strategies exist but only 4 (self-monitoring, goal-setting, rewards and social support) are usually included in weight loss applications. Studies have found the current implementations of these strategies are ineffective for weight-loss behavior change in majority of users. My work focuses on implementing and evaluating: (i) alternative rewards and social support strategies, and (ii) 2 additional strategies: stimulus control and time management. The results indicate users most prefer social support and rewards strategies implemented as working towards weight loss goals with a group of friends (team-based) for the former and a gamified approach with small gifts earned at unknown milestones for the latter. An initial evaluation on stimulus control indicates it is more effective than self-monitoring alone but a virtual-coaching element and just-in-time interventions are needed to engage users in the long-term. language of the presentation: English | ||
| 渡部 聡之 | 1461203: D, 中間発表 | 計算システムズ生物学 金谷 重彦, 笠原 正治, MD.ALTAF-UL-AMIN, 杉浦 忠男 |
| APICHANUKUL WORACHATE | 1551201: M, 1回目発表 | 大規模システム管理 笠原 正治 |
|
title: Accuracy Improvement for Speculative algorithms in Hadoop.
abstract: Hadoop, a well known distributed computing platform, is confronted with a problem called straggler problem that some tasks of a job take unusual long execution time and delay the completion time of the job. The slow tasks are known as straggler tasks. Hadoop uses a speculative algorithm which launches backup tasks. However, estimating the task execution time by existing algorithms is not so accurate that the speculative algorithm aggressively starts many backup tasks, half of which are unnecessary and wrong. To alleviate this problem, we propose an effective algorithm which has an elaborate estimation process to speculate the task execution time more precisely. In our algorithm, we separate the execution of a task into phases, and estimate the execution time of the task from the progress of the phases. Furthermore, we refine the estimation by combining the execution time information from other finished tasks and its own execution information, and by changing weights of the combination as time passes. Experimental results show that our proposed algorithm reduces not only 6.01% of numbers of unnecessary backup tasks, but also 28.69% of numbers of wrong backup tasks compared with the Hadoop's default algorithm. language of the presentation: English | ||