Faculty Profiles - Noboru Babaguchi

Papers - Noboru Babaguchi

Division display 41 - 60 of about 172 ／ All the affair displays >>

A Framework of Privacy-Preserving Image Recognition for Image-Based Information Services Reviewed

K. Fujii, K. Nakamura, N. Nitta, N. Babaguchi

MULTIMEDIA MODELING (MMM 2017), PT I 10132 40 - 52 2017

Nowadays mobile devices such as smartphones are widely used all over the world. Moreover, the performance of image recognition has dramatically increased by deep learning technologies. From these backgrounds, we think that the following scenario of information services could be realized in the near future: users take a photo and send it to a server, who recognizes the location in the photo and returns the users some information about the recognized location. However, this kind of client-server-based image recognition can cause a privacy issue because image recognition results are sometimes privacy sensitive. To tackle the privacy issue, in this paper, we propose a novel framework for privacy-preserving image recognition in which the server cannot uniquely identify the recognition result but users can do so. An overview of the proposed framework is as follows: First users extract a visual feature from their taken photo and transform it so that the server cannot uniquely identify the recognition result. Then users send the transformed feature to the server, who returns a candidate set of recognition results to the users. Finally, the users compare the candidates and the original visual feature for obtaining the final result. Our experimental results demonstrate the effectiveness of the proposed framework.

Effect of Junk Images on Inter-concept Distance Measurement: Positive or Negative? Reviewed

Y. Nagasawa, K. Nakamura, N. Nitta, N. Babaguchi

MULTIMEDIA MODELING, MMM 2017, PT II 10133 173 - 184 2017

In this paper, we focus on the problem of inter-concept distance measurement (ICDM), which is a task of computing the distance between two concepts. ICDM is generally achieved by constructing a visual model of each concept and calculating the dissimilarity score between two visual models. The process of visual concept modeling often suffers from the problem of junk images, i.e., the images whose visual content is not related to the given text-tags. Similarly, it is naively expected that junk images also give a negative effect on the performance of ICDM. On the other hand, junk images might be related to its text-tags in a certain (non-visual) sense because the text-tags are given by not automated systems but humans. Hence, the following question arises: Is the effect of junk images on the performance of ICDM positive or negative? In this paper, we aim to answer this non-trivial question from experimental aspects using a unified framework for ICDM and junk image detection. Surprisingly, our experimental result indicates that junk images give a positive effect on the performance of ICDM.

Tag Chat: A Tag-Based Past Topics Recollection Support System. Reviewed

J. Itou, R. Tanaka, J. Munemori, N. Babaguchi

Proc. of 9th International Conference on Collaboration Technologies (CollabTech 2017) LNCS 10397 29 - 36 2017

Long-term people reidentification using anthropometric signature Reviewed

M. Hasan, N. Babaguchi

2016 IEEE 8th International Conference on Biometrics Theory, Applications and Systems, BTAS 2016 2016.12

Anthropometric biometrics are the most suitable features for long-term people reidentification. However, feature selection is still a major problem in the anthropometric biometrics literature. In this paper, we aim at improving people reidentification by enhancing feature selection. Based on a statistical analysis of body measurements on a large-scale dataset, a new anthropometric signature is introduced. The proposed signature describes both the size and shape of human body at specific anatomical landmarks. While size is measured by Euclidian distance between four skeleton joint pairs, shape is described by the surface distance along four circular body parts. A novel algorithm is proposed to automatically segment the circular body parts from the subject point cloud using body geometry, cylindrical fitting and soft clustering. The overall system is evaluated on two public datasets using CMC and nAUC metrics. Experimental results show the effectiveness of our method compared to state of the art.

Hierarchical detection of rectangles in images Reviewed

M. Hasan, M. Abdellatif, N. Babaguchi

First International Workshop on Pattern Recognition 7 pages 2016.7

Privacy Protection for Social Video via Background Estimation and CRF-Based Videographer's Intention Modeling Reviewed

Yuta Nakashima, Noboru Babaguchi, Jianping Fan

IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS E99D ( 4 ) 1221 - 1233 2016.4

The recent popularization of social network services (SNSs), such as YouTube, Dailymotion, and Facebook, enables people to easily publish their personal videos taken with mobile cameras. However, at the same time, such popularity has raised a new problem: video privacy. In such social videos, the privacy of people, i.e., their appearances, must be protected, but naively obscuring all people might spoil the video content. To address this problem, we focus on videographers' capture intentions. In a social video, some persons are usually essential for the video content. They are intentionally captured by the videographers, called intentionally captured persons (ICPs), and the others are accidentally framed-in (non-ICPs). Videos containing the appearances of the non-ICPs might violate their privacy. In this paper, we developed a system called BEPS, which adopts a novel conditional random field (CRF)-based method for ICP detection, as well as a novel approach to obscure non-ICPs and preserve ICPs using background estimation. BEPS reduces the burden of manually obscuring the appearances of the non-ICPs before uploading the video to SNSs. Compared with conventional systems, the following are the main advantages of BEPS: (i) it maintains the video content, and (ii) it is immune to the failure of person detection; false positives in person detection do not violate privacy. Our experimental results successfully validated these two advantages.

Evaluating Protection Capability for Visual Privacy Information Reviewed

Y. Nakashima, T. Ikeno, N. Babaguchi

IEEE Security and Privacy 14 ( 1 ) 55 - 61 2016.1

Real-World Observation Extraction from Microblog based on Word Associative Relations Reviewed

Masato Yoshitake, Naoko Nitta, Noboru Babaguchi

2016 IEEE SECOND INTERNATIONAL CONFERENCE ON MULTIMEDIA BIG DATA (BIGMM) 450 - 455 2016

A large number of real-world observations by social sensors all over the world can be obtained from various social networking services. Especially, observations covering miscellaneous areas of interest are posted to Twitter as short text messages. Our goal is to extract a wide range of observations related to the target of interest specified by the user from Twitter regardless of their popularity. Assuming that the related observations are likely to contain words people often associate with each other, the associative relations among words are learned from the past messages. When a user gives a keyword representing his/her current interest, recent related observations are extracted based on their word frequency distributions in terms of the associative relations to the keyword.

Facial expression preserving privacy protection using image melding Reviewed

Yuta Nakashima, Tatsuya Koyama, Naokazu Yokoya, Noboru Babaguchi

Proceedings - IEEE International Conference on Multimedia and Expo 2015- 1 - 6 2015.8

An enormous number of images are currently shared through social networking services such as Facebook. These images usually contain appearance of people and may violate the people's privacy if they are published without permission from each person. To remedy this privacy concern, visual privacy protection, such as blurring, is applied to facial regions of people without permission. However, in addition to image quality degradation, this may spoil the context of the image: If some people are filtered while the others are not, missing facial expression makes comprehension of the image difficult. This paper proposes an image melding-based method that modifies facial regions in a visually unintrusive way with preserving facial expression. Our experimental results demonstrated that the proposed method can retain facial expression while protecting privacy.

Owner authentication for mobile devices using motion gestures based on multi-owner template update Reviewed

Shigeki Karita, Kumi Nakamura, Kazuhiro Kono, Yoshimichi Ito, Noboru Babaguchi

2015 IEEE International Conference on Multimedia and Expo Workshops, ICMEW 2015 2015.7

This paper proposes a template updating method for improving authentication accuracy in behavioral biometric authentication with hand/arm motion gestures for mobile devices. We introduce an extended version of the standard K-Medoids based clustering algorithm called supervised K-Medoids, which can handle with 2-class data such as positive samples and negative samples. Using the supervised K-Medoids, the template corresponding to each owner is selected as the one that is the most identifiable as the actual owner, and, at the same time, the most distinguishable from the others. Therefore, our method can decrease False-Rejection-Rate (FRR) and False-Acceptance-Rate (FAR) simultaneously, compared to the conventional work that is based on the template update with only the owner's data to decrease FRR. Our template update with multi-owner data attains Equal-Error-Rate (EER) of 5.2% whereas the conventional template update method with owner's own data results in 12.0% when 10 subjects authenticate with gestures for 10 days.

Temporal Spotting of Human Actions from Videos Containing Actor's Unintentional Motions Reviewed

K. Hara, K. Nakamura, N. Babaguchi

Proc. IEEE International Conference on Multimedia and Expo (ICME2015) 2015.7

Protection and Utilization of Privacy Information via Sensing Invited

Noboru Babaguchi, Yuta Nakashima

IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS E98D ( 1 ) 2 - 9 2015.1

Our society has been getting more privacy-sensitive. Diverse information is given by users to information and communications technology (ICT) systems such as IC cards benefiting them. The information is stored as so-called big data, and there is concern over privacy violation. Visual information such as images and videos is also considered privacy-sensitive. The growing deployment of surveillance cameras and social network services has caused a privacy problem of information given from various sensors. To protect privacy of subjects presented in visual information, their face or figure is processed by means of pixelization or blurring. As image analysis technologies have made considerable progress, many attempts to automatically process flexible privacy protection have been made since 2000, and utilization of privacy information under some restrictions has been taken into account in recent years. This paper addresses the recent progress of privacy protection for visual information, showing our research projects: PriSurv, Digital Diorama (DD), and Mobile Privacy Protection (MPP). Furthermore, we discuss Harmonized Information Field (HIFI) for appropriate utilization of protected privacy information in a specific area.

Real-Time People Counting across Spatially Adjacent Non-Overlapping Camera Views Reviewed

R. Akai, N. Nitta, N. Babaguchi

Proc. of International Conference on Multimedia Modeling (MMM2015) 71 - 82 2015.1

Digital diorama: Privacy-preserving and intelligible sensing-based real-world content Reviewed

Naoko Nitta, Noboru Babaguchi

ITE Transactions on Media Technology and Applications 3 ( 3 ) 184 - 193 2015

This paper proposes a sensing-based real-world content called Digital Diorama, which is a three-dimensional miniature model of the dynamically changing real world created from the real-time data continuously published over the Internet by stationary cameras distributed in public spaces. Digital Diorama is designed to preserve both the privacy of the monitored persons and the intelligibility of the content by superimposing the latest background images and human icons each of which represents a monitored person in the three-dimensional model. By using the data continuously published from 10 stationary cameras installed on one floor of a shopping mall, our prototype Digital Diorama browser was able to construct and dynamically update Digital Diorama in approximately 1 fps. The results of subjective evaluations indicated that utilizing appropriate human icons can improve the intelligibility of the content while preserving the privacy of the monitored persons.

Inter-Concept Distance Measurement with Adaptively Weighted Multiple Visual Features Reviewed

Kazuaki Nakamura, Noboru Babaguchi

COMPUTER VISION - ACCV 2014 WORKSHOPS, PT III 9010 56 - 70 2015

Most of the existing methods for measuring the inter-concept distance (ICD) between two concepts from their image instances use only a single kind of visual feature extracted from each instance. However, a single kind of feature is not enough for appropriately measuring ICDs due to a wide variety of perspectives for similarity evaluation (e.g., color, shape, size, hardness, heaviness, and functions); the relationships between different concept pairs are more appropriately modeled from different perspectives provided by multiple kinds of features. In this paper, we propose extracting two or more kinds of visual features from each image instance and measuring ICDs using these multiple features. Moreover, we present a method for adaptively weighting the visual features on the basis of their appropriateness for each concept pair. Experiments demonstrated that the proposed method outperformed a method using only a single kind of visual feature and one combining multiple kinds of features with a fixed weight.

Real-Time Local Word Database Construction from Twitter Reviewed

Takuya Kamimura, Naoko Nitta, Noboru Babaguchi

2015 IEEE INTERNATIONAL CONFERENCE ON SMART CITY/SOCIALCOM/SUSTAINCOM (SMARTCITY) 299 - 306 2015

Recently, geotagged posts to social media such as Twitter have been used to automatically construct a geographical dictionary containing diverse types of local words which indicate specific locations in the real world. The existing methods typically examine the spatial locality of the usage patterns observed in the geotagged posts accumulated for a certain period of time to select the local words; however, how long the geotagged posts need to be accumulated depends on the usage frequency of the word, and additionally, some local words can indicate different locations at different times. Thus, we propose a real-time method for constructing a local word database which consistently keeps the local words and their locations up to date by iteratively adding new local words, removing old temporary local words, and updating the locations indicated by the local words. These functions are realized by adaptively recording/resetting the usage history of each word to properly examine its spatial locality and by assigning the weight for each geotag which is used to represent the locations indicated by the local words according to their temporal variations. The local word database constructed by our proposed method was verified to contain more up-todate local words and locations compared to other types of geographical dictionary constructed by experts, crowdsourcing, and from the geotagged tweets accumulated for a fixed period of time based on the performance evaluations of tweet location estimation as an example of applications utilizing the geographical dictionaries.

Camera oscillation pattern for VSLAM: Translational versus rotational Reviewed

Mohamed Heshmat, Mohamed Abdellatif, Kazuaki Nakamura, A. A. Abouelsoud, Noboru Babaguchi

2014 International Conference on 3D Imaging, IC3D 2014 - Proceedings 2014

Visual SLAM algorithms exploit natural scene features to infer the camera motion and build a map of the environment landmarks. SLAM algorithm has two interrelated processes localization and mapping. For accurate localization, we need the features location estimates to converge quickly. On the other hand, to build an accurate map, we need accurate localization. Recently, a biologically inspired approach exploits deliberate camera oscillation has been used to improve the convergence speed of depth estimate. In this paper, we explore the effect of camera oscillation pattern on the accuracy of VSLAM. Two main oscillation patterns are used for distance estimation: translational and rotational. Experiments, using static and moving robot, are made to explore the effect of these oscillation patterns on the VSLAM performance.

Dynamic feature detection using virtual correction and camera oscillations Reviewed

Mohamed Heshmat, Mohamed Abdellatif, Kazuaki Nakamura, A. A. Abouelsoud, Noboru Babaguchi

2014 International Conference on 3D Imaging, IC3D 2014 - Proceedings 2014

Visual SLAM algorithms exploit natural scene features to infer the camera motion and build a map of a static environment. In this paper, we relax the severe assumption of a static scene to allow for the detection and deletion of dynamic points. A new 'virtual correction' method is introduced which serves to detect the dynamic points by checking the re-projection error of the points before and after the virtual measurement update. It can also recover the erroneously excluded useful features, particularly the distant points which may be deleted because of the change in its position after new measurement observation. Deliberate camera oscillations are also used to improve the VSLAM accuracy and the camera observability. The simulation results showed the effectiveness of the virtual correction when combined with camera oscillation in recovering the misclassified features and detecting the dynamic features even in difficult scenarios.

Real-world event detection using Flickr images Reviewed

Naoko Nitta, Yusuke Kumihashi, Tomochika Kato, Noboru Babaguchi

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 8326 ( 2 ) 307 - 314 2014

This paper proposes a real-world event detection method by using the time and location information and text tags attached to the images in Flickr. Events can generally be detected by extracting images captured at the events which are annotated with text tags frequently used only in specific times and locations. However, such approach can not detect events where only a small number of images were captured. We focus on the fact that semantically related events often occur around the same time at different locations. Considering a group of these events as an event class, the proposed method firstly detects event classes from all images in Flickr based on their similarity of the captured time and text tags. Then, from the images consisting each event class, events are detected based on their similarity of the captured locations. Such two-step approach enables us to detect events where a small number of images were captured. © 2014 Springer International Publishing.

Special Issue on Intelligent Video Surveillance for Public Security and Personal Privacy Invited

Noboru Babaguchi, Andrea Cavallaro, Rama Chellappa, Frederic Dufaux, Liang Wang

IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY 8 ( 10 ) 1559 - 1561 2013.10

　 PREV - NEXT 　

<Noboru Babaguchi>

Papers - Noboru Babaguchi