Satoshi Nakamura

Leveraging viewer comments for mood classification of music video clips

This short paper proposes a method to classify music video clips uploaded to a video sharing service into music mood categories such as cheerful, wistful, and aggressive. The method leverages viewer comments posted to the music video clips for the music mood classification. It extracts specific features from the comments: (1) adjectives in comments, (2) lengthened words in comments, and (3) comments in chorus sections. Our experimental results classifying 695 video clips into six mood categories showed that our method outperformed the baseline in terms of macro and micro averaged F-measures. In addition, our method outperformed the existing approaches that utilize lyrics and audio signals of songs.

VRMixer: mixing video and real world with video segmentation

This paper presents VRMixer, a system that mixes real world and a video clip letting a user enter the video clip and realize a virtual co-starring role with people appearing in the clip. Our system constructs a simple virtual space by allocating video frames and the people appearing in the clip within the users 3D space. By measuring the users 3D depth in real time, the time space of the video clip and the users 3D space become mixed. VRMixer automatically extracts human images from a video clip by using a video segmentation technique based on 3D graph cut segmentation that employs face detection to detach the human area from the background. A virtual 3D space (i.e., 2.5D space) is constructed by positioning the background in the back and the people in the front. In the video clip, the user can stand in front of or behind the people by using a depth camera. Real objects that are closer than the distance of the clips background will become part of the constructed virtual 3D space. This synthesis creates a new image in which the user appears to be a part of the video clip, or in which people in the clip appear to enter the real world. We aim to realize video reality, i.e., a mixture of reality and video clips using VRMixer.

A Visual Analytics Tool for System Logs Adopting Variable Recommendation and Feature-Based Filtering

Analysis and monitoring of system logs such as transaction logs and access logs is important for various objectives including trend discovery, update effort determination, and malicious behavior monitoring. However, it is not always an easy task because these logs may be massive, consisting of millions of records containing tens of variables, and therefore it may be difficult or time-consuming to discover significant knowledge. This paper presents a visual analytics tool which enables us to effectively observe system logs. The tool recommends variables that can reveal interesting discoveries and provides feature-based filtering that selects meaningful items from the visualization results. This paper also presents the result of experiments for non-professional users.

A sustainable practice method of hand-drawing by merging user's stroke and model's stroke

Nowadays, people have become able to publish their comics easily because of the spread of Web services featuring their own content. However, it is not easy for them to publish such comics because of the difficulty of improving their hand-drawing skill. One practice method is tracing. However, it is not easy for people to maintain motivation for this practice method. In this paper, we propose a method which enables people to improve their hand-drawing skills and motivation by creating a merging of their hand-drawn character and a corresponding model which is drawn by others automatically. We validate the usefulness of our method by using the system and show how to change the appearance of illustrations.

Deconaby: Animations for Improving Understandability of Web Images

There are many “how-to” pages on the Web. However, on such pages, descriptions and images are often not clearly linked. As a result, it is difficult for viewers to understand which parts of the text description refer to which pars of the images. This study aims to solve this problem. With the developed Deconaby system, users can add animations just by clicking on an image on a Web page and moving their mouse. Our system then generates a URL that consists of the how-to page’s original URL and a hash ID for sharing the animation-added version with others. In this manner, our system can improve the understandability of Web pages with little time and effort. We conducted an experiment to test the effectiveness of our system. The results showed that our system succeeded in increasing the understandability of difficult-to-understand pages. We also showed that a page written in an unfamiliar language can be understood with the help of appropriate Deconaby animations.

Image Blurring Method for Enhancing Digital Content Viewing Experience

Many systems have been studied for enhancing a user’s interest in digital content by using HMD, 3DTV, and so on. However, for these systems to be able to enhance a user’s interest, the creator needs to elaborate the content. In this paper, we present a method that extends the experience of digital content by simply superimposing blurring effects that follow the gaze point of the user. To clarify how the experience of viewing still images and videos was enhanced by our method, we compared user impressions when viewing digital content with and without our method. We also examined physiological impressions such as visibility and discomfort. The experimental results showed that the participants’ impressions of the video content were changed by superimposing the blurring effect on the peripherally viewed area. In particular, almost all psychological impression items (immersion, stereoscopic effect and so on) were scored higher when the blur was superimposed than when it was not.

Can People Sense Their Personalities Only by Watching the Movements of Their Skeleton in Street Dancing Performances

Dancing is a way of communication using the body and self-expression and is a kind of action where each individual’s uniqueness connects directly to his or her movements. Furthermore, how well a dancer can express his or her personality is one of the indicators of their ability, which proves the importance of personality in dancing. This study focused on personality in dancing, in particular the possibility of its extraction. Specifically, we asked skilled dancers and unskilled dancers to practice and perform a dance during which we acquired each individual’s bone structural data with the use of a Kinect sensor. Afterwards, we played back the data to each participant and asked them to choose what they thought were their own dancing forms, the form they most preferred, and the forms they thought were good. As a result, both the skilled and unskilled participants were capable of distinguishing their own dancing forms, which indicated the existence of a dancing personality. Furthermore, while there were differences between the skilled dancers and the unskilled ones, there was a common tendency of matching dance forms that participants favored and dance forms that participants considered good.

PhoToDo: image-based task management system by visual trigger

Many people manage their tasks using tools such as notebooks or personal task management applications in their smartphones. In fact, according to Microsofts research, 78% of respondents in the United States currently have at least one task management app [1]. However, conventional task lists are sometimes troublesome because tasks usually need to be expressed in words. In addition, it takes time to understand tasks when they are described in words. However, it is known that a person can instantaneously process an image and has the ability to process many images at once [2][3]. Therefore, we propose a system called PhoToDo that enables people to use visual images to manage tasks. By using PhoToDo, users can instantly visualize all their tasks and efficiently manage them. In this paper, we propose and implement our system and show its effectiveness by conducting experimental tests.

Mojirage: average handwritten note

Because some people appreciate their own handwritten characters blended with the handwritten characters of others, we propose a method for generating good handwriting by the real-time blending of users handwritten characters with their own past handwritten characters or others handwritten characters. We also realize the prototype system and show its usefulness.

Can Social Comments Contribute to Estimate Impression of Music Video Clips

The main objective of this paper is to estimate the impressions of music video clips using social comments to achieve impression-based music video clip searches or recommendation systems. To accomplish the objective, we generated a dataset that consisted of music video clips with evaluation scores on individual media and impression types. We then evaluated the precision with which each media and impression type were estimated by analyzing social comments. We also considered the possibility and limitations of using social comments to estimate impressions of content. As a result, we revealed that it is better to use proper parts-of-speech in social comments depending on each media/impression type.