Papers by Masamichi Shimosaka
Time-Series Human Motion Analysis with Kernels Derived from Learned Switching Linear Dynamics
Transactions of the Japanese Society for Artificial Intelligence, 2005
ABSTRACT In this paper, we propose a novel kernel computation algorithm between time-series human... more ABSTRACT In this paper, we propose a novel kernel computation algorithm between time-series human motion data for online action recognition. The proposed kernel is based on probabilistic models called switching linear dynamics (SLDs). SLD is one of the powerful tools for tracking, analyzing and classifying human complex time-series motion. The proposed kernel incorporates information about the latent variables in SLDs. The empirical evaluation using real motion data shows that a classifier using SVM with our proposed kernel has much better performance than the classifiers with some conventional kernel techniques. Another experimental result using kernel principal component analysis shows that the proposed kernel has excellent performance in extracting and separating different action categories, such as walking and running.
三次元ボクセルに基づく高速オンライン人体姿勢推定
Journal of the Robotics Society of Japan, 2008
ZigBee based wireless indoor localization with sensor placement optimization towards practical home sensing*
Advanced Robotics, 2016
Behavior prediction from trajectories in a house by estimating transition model using stay points
International Conference on Intelligent RObots and Systems - IROS, 2011
In this paper we propose a novel method for predicting resident's behaviors in a house from o... more In this paper we propose a novel method for predicting resident's behaviors in a house from one's move- ment trajectories. The method consists of 1) segmentation of trajectory data into staying or moving and classification of the segments and 2) prediction by time-series association rules from transition events of each segment. The method predicts the start time of target behaviors
Numerical modeling of image discriminability for home storage and organization system on a smart device
Proceedings of the 2013 ACM conference on Pervasive and ubiquitous computing adjunct publication - UbiComp '13 Adjunct, 2013
Hourly pedestrian population trends estimation using location data from smartphones dealing with temporal and spatial sparsity
Proceedings of the 22nd ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems - SIGSPATIAL '14, 2014
Human shape reconstruction via graph cuts for voxel-based markerless motion capture in intelligent environment
Proceedings of the 3rd International Universal Communication Symposium on - IUCS '09, 2009
In this paper, we propose a robust and real-time 3D human shape reconstruction method in daily li... more In this paper, we propose a robust and real-time 3D human shape reconstruction method in daily life spaces to make practical voxel-based motion capture systems. Our algorithm extracts human silhouette and reconstructs human shape via volume intersection from multi view point images. The method presented in this paper is based on energy minimization via graph cuts, and its main features are: 1) to reduce the background subtraction errors caused by background clutter, 2) to have robustness for influences of shadows, 3) to segment the foreground region even if moving objects other than human. The precise human shape reconstructed by the method improves the accuracy of human pose estimation. Especially, 3) leads to enhance the range of application of the voxel-based human pose estimation. We demonstrate the effectiveness of our approach in terms of both quantitative and qualitative performance where strong shadows appear and moving objects are present in intelligent environment.
Adaptive human shape reconstruction via 3D head tracking for motion capture in changing environment
2011 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2011
ABSTRACT This paper describes a human shape reconstruc- tion method from multiple cameras in dail... more ABSTRACT This paper describes a human shape reconstruc- tion method from multiple cameras in daily living environment, which leads to robust markerless motion capture. Due to contin- ual illumination changes in daily space, it had been difficult to get human shape by background subtraction methods. Recent statistical foreground segmentation techniques based on graph- cuts, which combine background subtraction information and image contrast, provide successful results; however, they fail to extract human shape when furniture such as tables and chairs are moved. In this paper, we focus on the results of face detectors that would be independent of such background changes and help to improve the robustness under movement of background objects. We propose a robust human shape reconstruction method with the following two characteristics. One is iterative image segmentation based on graph-cuts to integrate head position information into shape reconstruction. The other is high-precision head tracker to keep multi-view consistency. Experimental results show that proposed method has enhanced human pose estimation based on reconstructed human shape, and enables the system to deal with dynamic environment. extraction. The other is high-precision head tracker to keep multi-view consistency of head poses. There are many approaches to reconstruct human shape in spite of dynamic background. For example, voxel coloring which uses multi-view color consistency (4) and the fusion of multiple depth map acquired by stereo cameras (5) are able to reconstruct a target shape without silhouette extraction. Indeed, these reconstruction methods are not affected by background changes, but they take much more computational cost and need much more cameras. Moreover, it is necessary to extract human shape from whole shape of the target space through some additional clues. To achieve high-speed reconstruction, we employ silhouette-based approach, that is volume intersection. We tackle to extract human silhouette robustly with background and additional information. Re- cently, silhouette-based approaches with not only background information but also feedback from reconstruction results are proposed. Although feedback from reconstructed volumes (6) or feedback from estimated human pose (7) help us to extract human silhouette when background changes, these feedback are effective only when background changes are apart from human; furthermore, it is difficult to recover silhouette ex- traction if the feedback loop collapses. Our reconstruction method leverages head position estimated by textual features, which is independent of background information, to improve the robustness under movement of background objects.
Deploying RFID - Challenges, Solutions, and Open Issues, 2011
Development of wrist contour measuring device for an interface using hand shape recognition
Http Dx Doi Org 10 1080 01691864 2013 776939, Apr 5, 2013
ABSTRACT Recently, gesture recognition is widely used as interface. Popular gestures are mainly a... more ABSTRACT Recently, gesture recognition is widely used as interface. Popular gestures are mainly arm motion and whole body motion. Although hand shape is a good sign that can express rich information with small motions, few applications are in practical use. That is because the existing methods have several problems: blocks of finger sense and interference with finger motion, restrictions of hand position and posture, and complex initial configurations. In this study, we try to recognize hand shapes by observing the wrist contour, which varies with finger motions. We have developed a robust wrist-watch-type device that captures wrist contour, and have collected data from a substantial number of subjects. With the collected data, we conduct hand shape recognition experiments in several conditions. To overcome the positioning deviations and individual differences, two feature types are designed. Through the experiment, potential of the features is confirmed, and some effective features are picked up. In addition, concerning the design of recognition target properties, we examine the number of target hand shapes and the combination of hand shapes through the experiment, and several clues for target design are revealed.
Hand shape classification in various pronation angles using a wearable wrist contour sensor
Http Dx Doi Org 10 1080 01691864 2014 952337, Jan 26, 2015
ABSTRACT Hand gestures are potentially useful for communications between humans and between a hum... more ABSTRACT Hand gestures are potentially useful for communications between humans and between a human and a machine. However, existing methods entail several problems for practical use. We have proposed an approach to hand shape recognition based on wrist contour measurement. Especially in this paper, two assignments are addressed. First is the development of a new sensing device in which all elements are installed in a wrist-watch-type device. Second is the development of a new hand shape classifier that can accommodate pronation angle changes. The developed sensing device enables wrist contour data collection under conditions in which the pronation angle varies. The classifier recognizes the hand shape based on statistics produced through data forming and statistics conversion processes. The most important result is that no large difference exists between classification rates that include or those that exclude the independent (preliminary) pronation estimation process using inertia measurement units. This result shows two possible insights: (1) the wrist contour has some features that depend on the hand shape but which do not depend on the pronation angle, or (2) the wrist contour potentially includes pronation angle variation information. These insights indicate the possibility that hand shape can be recognized solely from the wrist contour, even while changing the pronation angle.
Predicting driving behavior using inverse reinforcement learning with multiple reward functions towards environmental diversity
2015 IEEE Intelligent Vehicles Symposium (IV), 2015
Fast Online Action Recognition with Boosted Combinational Motion Features
2006 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2006
ABSTRACT In this paper, we propose a fast and robust online action recognition method. The main f... more ABSTRACT In this paper, we propose a fast and robust online action recognition method. The main features of the proposed method are: 1) to select a small number of critical motion features from a very large set of motion feature templates and to release humans from task of designing critical motion features, 2) to require very small calculation cost for recognition compared to conventional methods, 3) to exploit "combinational motion features" which we propose as a new conception so as to construct a robust action recognizer. We evaluated the proposed method to gait action recognition, such as walking and running, by utilizing motion capture data. In the result, the proposed method reduced parameters given by human to action recognizer and lessened human's task. In addition, the proposed method needed very small calculation cost for recognition, and can recognize robustly as much as conventional action recognition method based on support vector machine. Moreover, the introduction of combinational motion features enhanced recognition performance
SVM-Based Human Action Recognition and Its Remarkable Motion Features Discovery Algorithm
Springer Tracts in Advanced Robotics, 2006
... The ultimate goal of their work is to recognize human speech and to be able to localize the p... more ... The ultimate goal of their work is to recognize human speech and to be able to localize the position of the human speaker. Page 2. ... First feature is Simultaneous Recognition. This is because human can recognize multiple action at the same time in parallel. ...
<title>Integrated virtual space control system utilizing hand gesture for intelligent house</title>
Sensor Fusion and Decentralized Control in Robotic Systems IV, 2001
ABSTRACT This paper proposes an integrated virtual space control system for intelligent house roo... more ABSTRACT This paper proposes an integrated virtual space control system for intelligent house room which consists of many kinds of appliances and information systems to enrich human life. The main features of the system are 1) the user controls the system via a large size display which represents system state linked to real appliances, 2) the system utilizes hand gesture to realize intuitive interaction, 3) the virtual space of the display utilizes 3D drag-and-drop metaphor. As for the core component of the proposed system, the paper reports implementation of a hand motion tracking system with multiple views and a controllable camera. Experimental results show that the tracking system solves a problem of the system scale explosion to achieve the accuracy and stability in real-time tracking.
Hierarchical recognition of daily human actions based on continuous Hidden Markov Models
Sixth IEEE International Conference on Automatic Face and Gesture Recognition, 2004. Proceedings., 2004
... ond is the problem of interpretation of the human motion, which includes modeling of action .... more ... ond is the problem of interpretation of the human motion, which includes modeling of action ... ing Hierarchical Structure and trust the recognition re-sult of the upper level, the ... data, continuous Hidden Markov Models and Feature Extraction Filter based on human expressions for ...
Marginalized Bags of Vectors Kernels on Switching Linear Dynamics for Online Action Recognition
Proceedings of the 2005 IEEE International Conference on Robotics and Automation, 2005
In this paper, we propose a novel kernel computation algorithm between time-series human motion d... more In this paper, we propose a novel kernel computation algorithm between time-series human motion data for online action recognition. The proposed kernel is based on probabilistic models called switching linear dynamics (SLDs). SLD is one of the powerful tools for tracking, analyzing and classifying human complex time-series motion. The proposed kernel incorporates information about the latent variables in SLDs with
Human Like Segmentation of Daily Actions based on Switching Model of Linear Dynamical Systems and Human Body Hierarchy
2006 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2006
ABSTRACT This paper presents a human like segmentation method for daily life actions, such as get... more ABSTRACT This paper presents a human like segmentation method for daily life actions, such as getting up, sitting down, walking. Unsupervised segmentation methods of many previous researches cannot always assure segmentation result that coincides with human&#39;s natural sense. While the proposed method utilizes human&#39;s teacher data of segmentation to conduct human like segmentation. We assume that latent dynamics changes at the segmentation points of action, and represent segmentation boundary by switching model of two linear dynamic systems. The problem is that human may segment actions according to wide variety of criteria depending on the attention point or other backgrounds. In this paper, those criteria are acquired by clustering segmentation boundaries extracted from teacher data made by human. Each of the cluster is characterized by body parts it pays attention to. Here, we focus on hierarchical aspect of human body that human body can be treated at various levels of abstraction (e.g. whole body, upper body, left arm), and represent it by tree structure. Experimental result shows that the proposed method can acquire human like segmentation criteria
Informative motion extractor for action recognition with kernel feature alignment
2004 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE Cat. No.04CH37566), 2004
ABSTRACT This paper proposes a novel algorithm for extracting informative motion features in dail... more ABSTRACT This paper proposes a novel algorithm for extracting informative motion features in daily life action recognition based on support vector machine (SVM). The main advantage of the proposed method is not only to extract remarkable motion features, which fit into human intuition, but also to improve the performance of the recognition system. Concretely speaking, the main properties of the proposed method are 1) optimizing kernel parameters so as to minimize its generalization error, 2) extracting remarkable motion features in response to the sensitivity of the kernel function. Experimental result shows that the proposed algorithm improves the accuracy of the recognition system and enables human to identify informative motion features intuitively.
Action recognition based on kernel machine encoding qualitative prior knowledge
2004 IEEE International Conference on Systems, Man and Cybernetics (IEEE Cat. No.04CH37583), 2004
ABSTRACT This paper proposes a recognition algorithm based on kernel classifier for human daily l... more ABSTRACT This paper proposes a recognition algorithm based on kernel classifier for human daily life action such as walking or lying down. The advantage of the proposed algorithm is to realize implant of qualitative human knowledge and robust recognition accuracy at the same time. The main features of the presented method are: (1)utilizing Gaussian process with latent variables for relation between recognized labels and input human motion, (2) in order to embed prior knowledge for proper recognition of novel motion dissimilar to the learned motion data, assigning probabilistic labels to virtual human motions generated in &quot;sparse&quot; area of input motion feature space, (3) learning parameters of classifier by real human motion with labels and the virtual motions in Bayesian perspective. The result of cross-validation like experiment shows that the accuracy of the proposed method is as good as support vector classification based recognition methods. It is also shown that the proposed method can recognize some novel motion fit into human common sense even when the classifiers without embedded knowledge fails to recognize it.
Uploads
Papers by Masamichi Shimosaka