Lab Meeting March 2, 2009 (Casey) : Nonrigid Structure from Motion in Trajectory Space

Title: Nonrigid Structure from Motion in Trajectory Space

Authors: Ijaz Akhter, Yaser Sheikh, SohaibKhan and Takeo Kanade

Abstract:

Existing approaches to nonrigid structure from motion assume that the instantaneous 3D shape of a deforming object is a linear combination of basis shapes, which have to be estimated anew for each video sequence. In contrast, we propose that the evolving 3D structure be described by a linear combination of basis trajectories. The principal advantage of this approach is that we do not need to estimate any basis vectors during computation. We show that generic bases over trajectories, such as the Discrete Cosine Transform (DCT) basis, can be used to compactly describe most real motions. This results in a significant reduction in unknowns, and corresponding stability in estimation. We report empirical performance, quantitatively using motion capture data and qualitatively on several video sequences exhibiting nonrigid motion including piece-wise rigid motion, partially nonrigid motion(such as a facial expression), and highly nonrigid motion(such as a person dancing). 

[Link]

Lab Meeting March 2, 2009(ZhenYu): Scale Invariant Feature Matching with Wide Angle Images

Title: Scale Invariant Feature Matching with Wide Angle Images(IROS 2007)
Authors: Hansen, P. Corke, P. Boles, W. Daniilidis, K.
Abstract:

Numerous scale-invariant feature matching algorithms using scale-space analysis have been proposed for use with perspective cameras, where scale-space is defined as convolution with a Gaussian. The contribution of this work is a method suitable for use with wide angle cameras. Given an input image, we map it to the unit sphere and obtain scale-space images by convolution with the solution of the spherical diffusion equation on the sphere which we implement in the spherical Fourier domain. Using such an approach, the scale-space response of a point in space is independent of its position on the image plane for a camera subject to pure rotation. Scale invariant features are then found as local extrema in scale-space. Given this set of scale-invariant features, we then generate feature descriptors by considering a circular support region defined on the sphere whose size is selected relative to the feature scale. We compare our method to a naive implementation of SIFT where the image is treated as perspective, where our results show an improvement in matching performance.

[LINK]

Automatic Class-Specific 3D Reconstruction from a Single Image

Title: Automatic Class-Specific 3D Reconstruction from a Single Image
Authors: Han-Pang Chiu, Leslie Pack Kaelbling, Tom´as Lozano-P´erez

Abstract: Our goal is to automatically reconstruct 3D objects from a single image, by using prior 3D shape models of classes. The shape models, defined as a collection of oriented primitive shapes centered at fixed 3D positions, can be learned from a few labeled images for each class. The 3D class model can then be used to estimate the 3D shape of an object instance, including occluded parts, from a single image. We provide a quantitative valuation of the shape estimation process on real objects and demonstrate its usefulness in three applications: robot manipulation, object detection, and generating 3D 'pop-up' models from photos.

Link: pdf

CMU talk: Scaling Up Game Theory: Representation and Reasoning with Action Graph Games

Intelligence Seminar

February 24, 2009

Scaling Up Game Theory: Representation and Reasoning with Action Graph Games
Kevin Leyton-Brown
Computer Science Department, University of British Columbia

Abstract:

Game theory is the mathematical study of interaction among independent, self-interested agents. It has wide applications, including the design of government auctions (e.g., for distressed securities), urban planning, and the analysis of internet traffic patterns. Interestingly, most work in game theory is analytic; it is less common to analyze a model's properties computationally. Key reasons for this are that game representation size tends to grow exponentially in the number of players--making all but the simplest games infeasible to write down--and that even when games can be represented, existing algorithms (e.g., for finding equilibria) tend to have worst-case performance exponential in the game's size.

This talk describes Action-Graph Games (AGG), which make it possible to extend computational analysis to games that were previously far too large to consider. I will give an overview of our five-year effort developing AGGs, emphasizing the twin threads of representational compactness and computational tractability. More specifically, the first part of the talk will describe the core ideas of the AGG representation. AGGs are a fully-expressive, graph-based representation that can compactly express both strict and context-specific independencies in players' utility functions. I will illustrate the representation by describing several practical examples of games that may be compactly represented as AGGs. The second part of the talk will examine algorithmic considerations. First, I'll describe a dynamic programming algorithm for computing a player's expected utility under a given mixed-strategy profile, which is tractable for bounded-in-degree AGGs. This algorithm can be leveraged to provide an exponential speedup in the computation of best response, Nash equilibrium, and correlated equilibrium. Second, I'll describe a message-passing algorithm for computing pure-strategy Nash equilibria in symmetric AGGs, which is tractable for graphs with bounded treewidth; again, this implies an exponential speedup over the previous state of the art. Finally, I'll more briefly describe some current directions in our work on AGGs: the modeling, evaluation, and comparison of different advertising auction designs; the extension of AGGs to both temporal and stochastic settings; and the design of free software tools to make it easier for other researchers to use AGGs.

This talk is based on joint work with Albert Xin Jiang, David R.M. Thompson, and Navin A.R. Bhat.

Bio:
Kevin Leyton-Brown is an assistant professor in computer science at the University of British Columbia. He received a B.Sc. from McMaster University (1998), and an M.Sc. and PhD from Stanford University (2001; 2003). Much of his work is at the intersection of computer science and microeconomics, addressing computational problems in economic contexts and incentive issues in multi-agent systems. He also studies the application of machine learning to the design and analysis of algorithms for solving hard computational problems. He has co-written two Multiagent Essentials of Game and over forty peer-refereed technical articles. He is an associate editor of the Journal of Artificial Intelligence Research (JAIR) and a member of the editorial board of the Artificial Intelligence Journal (AIJ). He has served as a consultant for Trading Dynamics Inc., Ariba Inc., and Cariocas Inc., and is currently scientific advisor to Worio Inc.

Lab Meeting February 23, 2009(Chung-Han) Human-Assisted Motion Annotation

Title : Human-Assisted Motion Annotation

Auther : Ce Liu, William T. Freeman, Edward H. Adelson, Yair Weiss (CSAIL MIT)

Abstract :
Obtaining ground-truth motion for arbitrary, real-worldvideo sequences is a challenging but important task for both algorithm evaluation and model design. Existing ground-truth databases are either synthetic, such as the Yosemite sequence, or limited to indoor, experimental setups, such as the database developed in [5]. We propose a human-in loop methodology to create a ground-truth motion database for the videos taken with ordinary cameras in both indoor and outdoor scenes, using the fact that human beings are experts at segmenting objects and inspecting the match between two frames. We designed an interactive computer vision system to allow a user to efficiently annotate motion.Our methodology is cross-validated by showing that human annotated motion is repeatable, consistent across annotators,and close to the ground truth obtained by [5]. Using our system, we collected and annotated 10 indoor andoutdoor real-world videos to form a ground-truth motion database. The source code, annotation tool and database is online for public evaluation and benchmarking.

Presented at CVPR 2008.

[Full Text]

Lab Meeting Feburary 23, 2009 (Jimmy): Multi-robot Cooperative Localization through Collaborative Visual Object Tracking

In: RoboCup 2007

Title: Multi-robot Cooperative Localization through Collaborative Visual Object Tracking

Authors: Zhibin Liu, Mingguo Zhao, Zongying Shi, and Wenli Xu

Abstract

In this paper we present an approach for a team of robots to cooperatively improve their self localization through collaboratively tracking a moving object. At first, we use a Bayes net model to describe the multi-robot self localization and object tracking problem. Then, by exploring the independencies between different parts of the joint state space of the complex system, we show how the posterior estimation of the joint state can be factorized and the moving object can serve as a bridge for information exchange between the robots for realizing cooperative localization. Based on this, a particle filtering method for the joint state estimation problem is proposed. And, finally, in order to improve computational efficiency and achieve real-time implementation, we present a method for decoupling and distributing the joint state estimation onto different robots. The approach has been implemented on our four-legged AIBO robots and tested through different scenarios in RoboCup domain showing that the performance of localization can indeed be improved significantly.

[link]

MIT talk: Do predictions of visual perception aid design?

Do predictions of visual perception aid design?
Speaker: Ruth Rosenholtz, MIT Brain and Cognitive Sciences
Date: Friday, February 20 2009
Time: 2:00PM to 3:00PM

Understanding and exploiting the abilities of the human visual system is an important part of the design of usable user interfaces and information visualizations. Designers traditionally learn qualitative rules-of-thumb for how to enable quick, easy and veridical perception of their design. More recently, work in human and computer vision, including in our lab, has produced more quantitative models of human perception. These models often take as input arbitrary, complex images of a design. We ask whether such models aid the design process. Through a series of interactions with designers and design teams, we find that the models can help, but in somewhat unexpected ways. Based on this study, I will suggest general design principles for perceptual tools.

Ruth Rosenholtz is a Principal Research Scientist in MIT's Department of Brain and Cognitive Sciences, and a member of CSAIL. She joined MIT in 2004 after 7 years at the Palo Alto Research Center (formerly Xerox PARC). Ruth's background is in electrical engineering, particularly computer vision. More recently, however, she has studied human vision, and in particular visual search and attention. Her engineering background shows through in her focus on finding predictive mathematical models of phenomena in human vision, and in her interest in applying knowledge of human perception to the design of better user interfaces and information visualizations.

Honda Research Institute : Summer internship 2009

The computer science research section of Honda Research Institute USA, located in Mountain View, California and Cambridge, Massachusetts, has several openings for internships this summer. For details, please go to:

http://www.honda-ri.com/HRI_Us/careers/internships

The application deadline is March 31, 2009.

CMU talk: Bringing people in the loop

VASC Seminar
Monday, February 16

Bringing people in the loop: data annotation and real-time visual supervision with Amazon Mechanical Turk
Alexander Sorokin
University of Illinois at Urbana-Champaign


Abstract:

In this talk I will challenge the common assumption that "we will never have enough labeled data". As of today, large annotated datasets can be built very quickly and cheaply. To achieve that a number of questions needs to be answered. I will discuss what properties of the tasks affect the quality of the final result as well as cost and speed of task completion. I will further highlight how people in the loop framework expands the range of possible applications.

Bio:
Alexander Sorokin is a 5th year PhD student at the University of Illinois at Urbana-Champaign. He has received BS from Lomonosov Moscow State University in 2003. In 2007 he received a best paper award at InfoVis 2007 for the paper "Visualizing the History of Living Spaces".

CMU talk: Large Scale Scene Matching for Graphics and Vision

Speaker: James Hays (CSD@CMU)
Date: Monday, Feb 16, 2009
Time: 12:00 noon

Title:
Large Scale Scene Matching for Graphics and Vision

Abstract:
The complexity of the visual world makes it difficult for computer vision to understand images and for computer graphics to synthesize visual content. The traditional computer vision or computer graphics pipeline mitigates this complexity with a bottom up, divide and conquer strategy (e.g. segmenting then classifying, assembling part-based models, or using scanning window detectors). In this talk I will discuss research that is fundamentally different, enabled by the observation that while the space of images is infinite, the space of "scenes" might not be astronomically large. With access to imagery on an Internet scale, for most images there exist numerous semantically and structurally similar scenes. My research is focused on exploiting and refining large scale scene matching to short circuit the typical computer vision and graphics pipelines for tasks such as scene completion, image geolocation, object detection, and high interest photo selection.

CMU talk: Intelligent Preference Assessment: The Next Steps?

Intelligence Seminar

February 3, 2009
3:30 pm


Intelligent Preference Assessment: The Next Steps?

Craig Boutilier, Department of Computer Science, University of Toronto

Preference elicitation is generally required when making or recommending decisions on behalf of users whose utility function is not known with certainty. Full elicitation of user utility functions is infeasible in practice, leading to an emphasis on approaches that (a) attempt to make good recommendations with incomplete utility information; and (b) heuristically minimize the amount of user interaction needed to assess relevant aspects of a utility function. Current techniques are, however, limited in a number of ways: (i) they rely on specific forms of information for assessment; (ii) they require very stylized forms of interaction; (iii) they are limited in the types of decision problems that can be handled.

In this talk, I will outline several key research challenges in taking preference assessment to a point where wide user acceptance is possible. I will focus on three three current techniques we're developing that will help move in the direction of greater user acceptance. Each tackles one of the weaknesses discussed above.

1. The first two techniques allows users to define "personalized" features over which they can express their preferences. Users provide (positive and negative) instances of a concept (or feature) over which they have preferences. We relate this to models of concept learning, and discuss how they existence of utility functions allows decisions to be made with very incomplete knowledge of the target concept. I'll also discuss possible means integrating data-intensive collaborative filtering approaches with explicit preference elicitation techniques, especially when tackling "subjective" features.

2. I'll discuss some of our recent work on applying explicit decision-theoretic models to more "conversational" critiquing approaches to recommender systems. We consider several semantics (wrt user preferences) for unstructured user choices and show how these can be integrated into regret-based models.

3. Time permitting, I'll provide a sketch of some recent work on eliciting reward functions in Markov decision processes using the notion of minimax regret.

Bio:

Craig Boutilier received his Ph.D. in Computer Science (1992) from the University of Toronto, Canada. He is Professor and Chair of the Department of Computer Science at the University of Toronto. He was previously an Associate Professor at the University of British Columbia, a consulting professor at Stanford University, and a visiting professor at Brown University. He has served on the Technical Advisory Board of CombineNet, Inc. since 2001.

Dr. Boutilier's research interests span a wide range of topics, with a focus on decision making under uncertainty, including preference elicitation, mechanism design, game theory, Markov decision processes, and reinforcement learning. He is a Fellow of the American Association of Artificial Intelligence (AAAI) and the recipient of the Isaac Walton Killam Research Fellowship, an IBM Faculty Award and the Killam Teaching Award. He has also served in a variety of conference organization and editorial positions, and is Program Chair of the upcoming Twenty-first International Joint Conference on Artificial Intelligence (IJCAI-09).

CMU talk: Perception on an Offroad Robot: Shallow and Deep Learning Architectures

VASC Seminar
Monday, February 2, 2009

Perception on an Offroad Robot: Shallow and Deep Learning Architectures
Raia Hadsell
Robotics Institute, CMU

Abstract:
Perception for offroad mobile robots is very difficult. Roads and paths aren't visually consistent, nor are they guaranteed to exist at all; obstacles are diverse and often visually complex. In addition, long range perception is more important when path planning needs to be done in the absence of clear roads or corridors. I will describe 2 learning-based approaches to perception in offroad environments: first, a kernel-based method for rough terrain reconstruction, and second, a self-supervised online vision system that can detect paths and obstacles at very long range and that quickly adapts to new environments.

Raia Hadsell is currently a postdoc at the Robotics Institute, working with Martial Hebert, Drew Bagnell, and Daniel Huber. She completed her doctorate at New York University in 2008, under the advisement of Yann LeCun, with research interests that lie in the intersection of machine learning, vision, and robotics. She has also enjoyed internships at Google NYC and Net-Scale Technologies. Dr. Hadsell, if pressed, will own up to her bachelor's degree in religion and philosophy and may discuss Nietzsche on occasion.

Lab Meeting January 19, 2009 (Tiffany): Algorithms for Inverse Reinforcement Learning

ICML 2000

Title: Algorithms for Inverse Reinforcement Learning

Authors: Andrew Y. Ng and Stuart Russell

Abstract:

This paper addresses the problem of inverse reinforcement learning (IRL) in Markov decision processes, that is, the problem of extracting a reward function given observed, optimal behaviour. IRL may be useful for apprenticeship learning to acquire skilled behaviour, and for ascertaining the reward function being optimized by a natural system. We rst characterize the set of all reward functions for which a given policy is optimal. We then derive three algorithms for IRL. The rst two deal with the case where the entire policy is known; we handle tabulated reward functions on a nite state space and linear functional approximation of the reward function over a potentially in-nite state space. The third algorithm deals with the more realistic case in which the policy is known only through a nite set of observed trajectories. In all cases, a key issue is degeneracy|the existence of a large set of reward functions for which the observed policy is optimal. To remove degeneracy, we suggest some natural heuristics that attempt to pick a reward function that maximally dierentiates the observed policy from other, suboptimal policies. This results in an eciently solvable linear programming formulation of the IRL problem. We demonstrate our algorithms on simple discrete / nite and continuous /in nite state problems.

link

Lab Meeting January 19, 2009 (Yu-chun): Interaction with a Zoomorphic Robot that Exhibits Canid Mechanisms of Behaviour

ICRA 2008

Title: Interaction with a Zoomorphic Robot that Exhibits Canid Mechanisms of Behaviour

Authors: Trevor Jones1, Shaun Lawson, and Daniel Mills

Abstract:

Despite parallels between the cooperative use of domestic dogs in human society today, the predicted similar deployment of robots in the future, and the plethora of superficially dog-like robotic entertainment devices, very little effort has been directed at exploiting any understanding of social cognition between dogs and humans when designing interactive robotic systems. This paper describes an experiment in which we gave interactive robots zoomorphic appearances and dog-like behavioural properties. We analysed human reactions to robots exhibiting differing levels of zoomorphism and dog-like behaviour during an interaction task; we were particularly interested to determine whether behaviour and/or appearance that mimicked that of dogs facilitated increased satisfaction in robot performance and a willingness to persevere with a robot that made mistakes. Our findings show that neither the appearance or behaviour of a robot had an impact on the participants’ rating of robot performance whilst there was also no significant difference in the self-reported categories of frustration, excitement and desire to persist with an interaction. However, our findings suggest that differences in individual preferences are revealed when people are asked to interact with robots that exhibit dog-like behaviours and other zoomorphic characteristics and that further research is required in order to better understand these differences.

link

CMU RI Thesis Proposal: Distributed Algorithms for Probabilistic Inference and Learning

Date: 16 January 2009

Time: 12:00 p.m.

Place: Newell Simon Hall 1507

Type: Thesis Proposal

Who:  Stanislav Funiak

Topic: Distributed Algorithms for Probabilistic Inference and Learning

Abstract:

Probabilistic inference and learning problems arise naturally in distributed systems such as sensor networks, teams of mobile robots, and recommendation systems. In these systems, the data resides at multiple distributed locations, and the network nodes need to collaborate, in order to perform the inference or learning task.

This thesis has three thrusts. First, we propose distributed implementations of several state-of-the-art centralized inference algorithms. Our solutions address challenges, such as effective MAP estimation, scheduling of messages in loopy belief propagation, and assumed density filtering.

Many algorithms for probabilistic inference are described by graphical models, such region graphs or junction trees. These graphical models, together with the update schedule, entirely determine the behavior of the inference algorithm in a centralized settings. Yet, in distributed settings, the graphical model crucially interacts with the physical network and determines properties, such as robustness or communication complexity. In this thesis, we propose a unified view where the graphical model and its placement is optimized jointly to match both the network and the probabilistic model. In this manner, our distributed algorithms will not only attain accurate solutions, but will also have a low message complexity.

Recent advances in peer-to-peer networks offer interesting opportunities for learning latent variable models for collaborative filtering. Peer-to-peer networks simplify many aspects of distributed learning, but open an interesting challenge of supporting recommendation queries with stale local models. We propose a pull-based approach that updates the model parameters, in order to minimize its regret with respect to the optimal set of recommendations.

We demonstrate our algorithms on real-world applications in large-scale modular robot localization, camera networks and movie recommendation systems. We demonstrate that our algorithms scale to large networks and provide improved robustness and convergence properties.

Full Text - http://www.cs.cmu.edu/~sfuniak/pubs/proposal.pdf

NTU talk (this Saturday): Non-Chronological Video Editing and Video Synopsis

Speaker: Prof. Shmuel Peleg, School of Computer Science and Engineering, The Hebrew University of Jerusalem, Israel

Time: 02:20pm, January 10 (Saturday), 2009

Place: Room 101, CSIE building

Title: Non-Chronological Video Editing and Video Synopsis

Abstract:
Powerful effects in video editing can be obtained when relaxing the chronological constraints: activities that occurred in different times can be shown simultaneously and vice versa. The description of non-chronological video editing effects and the simple methods to perform them will start this talk.

The non-chronological approach to video is also powerful in creating video summaries. In particular, a full day recorded by a video surveillance camera can be summarized in a few minutes without loss of any activity. It is estimated that 40 million surveillance cameras are being installed annually. But none of the video they record is ever watched: it is too time consuming. The presented video synopsis approach can provide access to the untapped resource of recorded surveillance cameras.

Short Biography:
Shmuel Peleg received his Ph.D. from the University of Maryland in 1979 under the guidance of Professor Azriel Rosenfeld. In 1981 he became a faculty member at the Hebrew University of Jerusalem where he is still a Professor of Computer Science. Shmuel served as chairman of the Institute of Computer Science at Hebrew University from 1990 to 1993.

Shmuel's research covers pyramid representation, image enhancement, motion analysis, panoramic mosaicing, and video surveillance. He has several patents which provided the technical foundations to four start-up companies. The most recent company, BriefCam Ltd., provides indexing into video surveillance, and uses a technology covered in this presentation.

NTU talk: Lecture Series on Brain Theory and Neural Network

Speaker：Dr. Michael A. Arbib,
Professor of Biomedical Engineering, Computer Science, Electrical Engineering Neurobiology, and Psychology, University of Southern California

Lecture Series on Brain Theory and Neural Network (I)
Time：2009/01/10 (Saturday) 9:00~12:00
Title: An introduction of Brain theory and artificial intelligence:
(1)Brief overview of AI & BT networks of leaky integrator neurons
(2)Winner-Take-All(a frog model);Didday-Arbib & Itti-Koch on visual attention
Place：台大生命科學館3樓視聽教室

Lecture Series on Brain Theory and Neural Network (II)
Time：2009/01/11 (Sunday) 9:00~12:00
Topics：(1)Dominey Arbib model of perception and attention:Working Memory & Dynamic Remapping
(2)Object Recognition & Scene Perception
Place：台大生命科學館3樓視聽教室

Lecture Series on Brain Theory and Neural Network(III)
Time: 2009/01/12(Monday) 9:00~12:00
Tpoics：Adaptive networks; reinforcement learning
(1)Introduction to Hebbian, supervised and reinforcement learning in neural networks
(2)Augmented Competitive Queuing: Opportunistic scheduling with mirror neurons and reinforcement learning
Place：台大生命科學館3樓視聽教室

NTU talk: Toward Robust Online Visual Tracking

Speaker: Prof. Ming-Hsuan Yang, UC Merced

Time: 02:20pm, January 9 (Fri), 2009
Place: Room 102, CSIE building

Title: Toward Robust Online Visual Tracking

Abstract:

Human beings are capable of tracking objects in dynamic scenes effortlessly, and yet visual tracking remains a challenging problem in computer vision. The main reason can be attributed to the difficulty in handling appearance variation of a target object. Intrinsic appearance change include out-of-plane motion and shape deformation of a target object, whereas extrinsic illumination change, camera motion, camera viewpoint, and occlusions inevitably cause large appearance variation.

Visual tracking is a fundamental problem in computer vision that has important applications in a variety of areas, including recovering 3D structure from moving scenes, camera calibration, estimating the underlying motion of the scene, and object recognition. It also has other applications in autonomous robotics and vehicles, medical imaging, as well as entertainment. While existing algorithms are able to track objects in controlled environments, they usually fail in the presence of significant image variations caused by changes in illumination, pose and occlusions. In addition, most of them require significant efforts in offline training prior to tracking. In the first part of this talk, I will present an efficient online learning algorithm for simultaneously tracking objects and learning compact appearance models. Numerous experiments show that this method is able to learn compact generative models for tracking target objects undergoing large pose and
illumination changes. I will then discuss discriminative algorithms that track objects by separating foreground targets from backgrounds in an online manner. Experimental validation demonstrates that these algorithms are robust for tracking fast moving objects undergoing illumination change, occlusion, and articulated motion in real time with better results than existing systems.


Short Biography:

Ming-Hsuan Yang is an assistant professor in Electrical Engineering and Computer Science of University of California at Merced. After receiving his PhD degree in Computer Science from the University of Illinois at Urbana-Champaign (UIUC), he worked as a senior researcher at the Honda Research Institute in Mountain View, California, and was an assistant professor with Computer Science and Information Engineering at National Taiwan University.

His research interests include computer vision, pattern recognition, robotics, cognitive science, and machine learning. While at UIUC, he was awarded the Ray Ozzie Fellowship given to outstanding graduate students in Computer Science. He has co-authored the book Face Detection and Gesture Recognition for Human-Computer Interaction (Kluwer Academic Publishers), and co-edited a special issue on face recognition of Computer Vision and Image Understanding. He serves as an Associate Editor of the IEEE Transactions on Pattern Analysis and Machine Intelligence, and an Area Chair of the IEEE Computer Vision and Pattern Recognition in 2008 and 2009. He is a senior member of the IEEE and the ACM.

CMU talk: Shape Constrained Figure-Ground Segmentation and Tracking

Special VASC Seminar
Thursday, January 8, 2009

Shape Constrained Figure-Ground Segmentation and Tracking
Zhaozheng Yin
Penn State University

Abstract:
To avoid drift problems during adaptive tracking, we must raise the level of abstraction at which the tracker represents its target. The goal must tracking "objects", not a box of pixels or a color distribution. If we can explicitly segment the foreground from background, it is possible to keep the adaptive model anchored on just the foreground pixels. In this talk, we discuss a shape constrained segmentation approach for tracking. Global object shape information is embedded into local graph links in a Conditional Random Field framework, thus the graph cut is attracted to occur around the figure-ground boundary. When treating tracking as a figure-ground segmentation problem, the precise foreground matte can help reduce pixel classification error during model adaptation. Meanwhile, the collected shape templates are useful to search for and recognize the same object after occlusion or tracking failure.

Bio:
Zhaozheng Yin is currently a PhD candidate in Robert Collins's vision group at Penn State, where his research interests include object segmentation, tracking, motion detection and feature selection/fusion. He received his BS degree from Tsinghua University, China, and his MS degree from the University of Wisconsin at Madison.

[NIPS 2008]Nonrigid Structure from Motion in Trajectory Space

Title: Nonrigid Structure from Motion in Trajectory Space

Authors: Ijaz Akhter, Yaser Ajmal Sheikh, Sohaib Khan, and Takeo Kanade
Published in: Neural Information Processing Systems (NIPS), December 2008

Abstract:
Existing approaches to nonrigid structure from motion assume that the instantaneous 3D shape of a deforming object is a linear combination of basis shapes, which have to be estimated anew for each video sequence. In contrast, we propose that the evolving 3D structure be described by a linear combination of basis trajectories. The principal advantage of this approach is that we do not need to estimate any basis vectors during computation. We show that generic bases over trajectories, such as the Discrete Cosine Transform (DCT) basis, can be used to compactly describe most real motions. This results in a significant reduction in unknowns, and corresponding stability in estimation. We report empirical performance, quantitatively using motion capture data, and qualitatively on several video sequences exhibiting nonrigid motions including piece-wise rigid motion, partially nonrigid motion (such as a facial expression), and highly nonrigid motion (such as a person dancing).

[Link to the paper and the datasets]