CGI Seminar Series at I2R
Seminar 25
Title: Design of an Intelligible Mobile Context-Aware Application
Speaker: Brian Lim, Human-Computer Interaction Institute, CMU
Chaired by: Jamie Ng Suat Ling
Venue: Franklin, 11 South, Connexis,
Fusionopolis
Time: 15:00-15:45, January 5, Thursday, 2012
Abstract
Context-aware
applications are increasingly complex and autonomous, and research has
indicated that explanations can help users better understand and ultimately
trust their autonomous behavior. However, it is still
unclear how to effectively present and provide these explanations. This work
builds on previous work to make context-aware applications intelligible by
supporting a suite of explanations using eight question types (e.g., Why, Why
Not, What If). We present a formative study on design and usability issues for
making an intelligible real-world, mobile context-aware application, focusing
on the use of intelligibility for the mobile contexts of availability, place,
motion, and sound activity. We discuss design strategies that we considered,
findings of explanation use, and design recommendations to make intelligibility
more usable.
Bio-data
Brian Lim is a 5th
year Ph.D. student at the Human-Computer Interaction Institute in Carnegie
Mellon University. He has been working with Prof. Anind
Dey to investigate how to make context-aware
applications intelligible, so that they can explain how they function, and
users would find their behavior more believable, and
find them more usable. He received his B.S. from Cornell University under the
A*STAR National Science Scholarship (BS) and is currently funded by the A*STAR
National Science Scholarship (PhD). Previously he was a research officer with I2R
from 2006-2007.
Seminar 24
Title: Evaluating Gesture-based Games with Older Adults on a Large Screen Display
Speaker: Dr Mark David Rice
Chaired by: Jamie Ng Suat Ling
Venue: Franklin, 11 South, Connexis,
Fusionopolis
Time: 10:00-10:30, July 6, Wednesday, 2011
Abstract
Focusing on
designing games for healthy older adults, we present a study exploring the usability
and acceptability of a set of three gesture-based games. Designed for a large
projection screen display, these games employ vision-based techniques that center on physical embodied interaction using a graphical
silhouette. Infrared detection, accompanied by back-projection is used to
reduce the effects of occluded body movements. User evaluations with 36 older
adults were analyzed using a combination of pre- and post-game questionnaires,
direct observations and semi-structured group interviews. The results
demonstrate that while all the games were usable, they varied in their physical
and social engagement, perceived ease of use and perceived usefulness. In
particular, items associated with physical wellbeing were rated highly.
Bio-data
Mark is a member of
the Human Factors Engineering Programme in the CGI department. Previously, he
was employed at the University of Dundee (UK) where he
graduated with a Ph.D. in Computer Science in 2009. In 2002 he joined the
Interactive Technologies Research Group at the University of
Brighton (UK), and in 2005 completed a research placement at the Advanced
Telecommunication Research Institute (Japan). During his research career, Mark
has worked on funded projects by the Alzheimer’s Association (US) and Intel
Corporation, and the European Commission (FP7). He has co-organized two
international workshops on inclusive design at EuroITV
2008 and Interact 2009, and has been a board member of Scotland’s largest
disability organization
- Capability Scotland.
Seminar 23
Title: Intrinsic Images Decomposition Using a Local and Global Sparse Representation of Reflectance
Speaker: Dr Shen Li
Chaired by: Dr Huang Zhiyong
Venue: Franklin, 11 South, Connexis,
Fusionopolis
Time: 10:00-11:00, May 5, Thursday, 2011
Abstract
Intrinsic image
decomposition is an important problem that targets the recovery of shading and
reflectance components from a single image. While this is an ill-posed problem
on its own, we propose a novel approach for intrinsic image decomposition using
a reflectance sparsity prior that we have developed.
Our method is based on a simple observation: neighboring
pixels usually have the same reflectance if their chromaticities
are the same or very similar.
We formalize this sparsity constraint on local reflectance, and derive a
sparse representation of reflectance components using data-driven
edge-avoiding-wavelets. We show that the reflectance component of natural
images is sparse in this representation. We also propose and formulate a novel
global reflectance sparsity constraint. Using the sparsity prior and global constraints, we formulate a l1-regularized least squares minimization problem for intrinsic
image decomposition that can be solved efficiently. Our algorithm can
successfully extract intrinsic images from a single image, without using other
reflection or color models or any user interaction.
The results on challenging scenes demonstrate the power of the proposed
technique.
Bio-data
Shen
Li received her PhD 2006 and M.Eng 2002, both in
computer science from Osaka University, Japan. Before joining I2R, she worked
as researcher at Microsoft Research Asia, and ECE Department in National University
of Singapore. Her main research interests are in computer graphics &
compute vision, especially in reflectance and illumination modeling,
image-based rendering/modeling and low-level vision.
Seminar 22
Title: Projected Interfaces and Virtual Characters
Speaker: Dr Jochen Walter Ehnes
Chaired by: Dr Corey Manson Manders
Venue: Franklin, 11 South, Connexis,
Fusionopolis
Time: 14:00-15:00, April 7, Thursday, 2011
Abstract
During
this seminar I will provide an overview of my previous work with computer
graphics, and 3D user interfaces at Fraunhofer IGD,
Tokyo University, as well as the University of Edinburgh. In particular I will
describe what motivated me to start working with projected displays.
Furthermore,
I would like to start a discussion about how my previous work with controllable
projectors could be turned into a project here at A*Star. I certainly see a lot
of possibilities for collaborations, so I am looking forward to an interesting
discussion.
Bio-data
Before
joining I2R, Jochen worked for
four years at the School of Informatics at the University of Edinburgh. While
he was a member of the Centre for Speech Technology Research (CSTR), he did
work on the graphical side of multimedia interfaces, complementing speech
technology developed by his colleagues. He also collaborated with people at the
Edinburgh College of Art.
Originally from
Germany, he obtained his Diplom (German equivalent to
the masters) in Informatics at the Technical University of Darmstadt in 2000.
During his studies he already worked part time at the Fraunhofer
Institute of Computer Graphics (Fraunhofer IGD). He
did his masters project at CAMTech, a shared lab
between Fraunhofer IGD and NTU at NTU. After that he
worked full time at IGD for two and a half years before he went to Tokyo to
work on his PhD, which he obtained in 2006.
Seminar 21
Title: Real-Time Data Driven Deformation Using Kernel Canonical Correlation Analysis
Speaker: Dr Kim Byung Uck
Chaired by: Dr Ishtiaq Rasool Khan
Venue: Franklin, 11 South, Connexis,
Fusionopolis
Time: 14:00-15:00, April 1, Friday, 2011
Abstract
Achieving intuitive
control of animated surface deformation while observing a specific style is an
important but challenging task in computer graphics.
Solutions to this task can find many applications in data-driven skin
animation, computer puppetry, and computer games. In this paper, we present an
intuitive and powerful animation interface to simultaneously control the
deformation of a large number of local regions on a deformable surface with a
minimal number of control points. Our method learns suitable
deformation subspaces from training examples, and generate new
deformations on the fly according to the movements of the control points. Our
contributions include a novel deformation regression method based on kernel
Canonical Correlation Analysis (CCA) and a Poisson-based translation solving
technique for easy and fast deformation control based on examples. Our run-time
algorithm can be implemented on GPUs and can achieve a few hundred frames per
second even for large datasets with hundreds of training examples.
Bio-data
Byung-Uck
Kim is a Senior Research Fellow in the Department of Computer Graphics &
Interface (CGI) at I2R. He is a voting member of ACM SIGGRAPH Singapore Chapter
and serves as an ACM SIGGRAPH reviewer. He received his B.S. (1996), M.S(1995), and Ph.D.(2004) from Yonsei
University, Seoul, Korea. He worked for Samsung Electronics Co. Ltd as a Senior
Engineer until 2007 and was a Postdoctoral Research Associate in the Department
of Computer Science at the University of Illinois, Urbana-Champaign until 2010.
Seminar 20
Title: Smoothly Varying Affine stitching
Speaker: Daniel Lin Wenyen
Chaired by: Dr Ng Tian Tsong
Venue: Franklin, 11 South, Connexis,
Fusionopolis
Time: 14:00-15:00, March 24, Thursday, 2011
Abstract
We design a new image stitching algorithm.
Traditional image stitching using parametric transforms such as homography, only produces perceptually correct composites for
planar scenes or parallax free camera motion between source frames. This limits
mosaicing to source images taken from the same
physical location. In this paper, we introduce a smoothly varying affine
stitching field which is flexible enough to handle parallax while retaining the
good extrapolation and occlusion handling properties of parametric transforms.
Our algorithm which jointly estimates both the stitching field and correspondence, permits the stitching of general motion
source images, provided the scenes do not contain abrupt protrusions.
Bio-data
Daniel Lin
undertook his undergraduate and PhD studies at the National University of
Singapore. His research interest is structure from motion and large
displacement matching.
Seminar 19
Title: Toolkit to Support Intelligibility in Context-Aware Applications
Speaker: Brian Lim, Human-Computer Interaction Institute, CMU
Chaired by: Tan Yiling Odelia
Venue: Transform 1 & 2 @ Level 13, North, Connexis, Fusionopolis
Time: 14:00-15:00, December 21, Tuesday, 2010
Abstract
Context-aware applications should be
intelligible so users can better understand how they work and improve their
trust in them. However, providing intelligibility is non-trivial and requires
the developer to understand how to generate explanations from application
decision models. Furthermore, users need different types of explanations and
this complicates the implementation of intelligibility. We have developed the
Intelligibility Toolkit that makes it easy for application developers to obtain
eight types of explanations from the most popular decision models of
context-aware applications. We describe its extensible architecture, and the
explanation generation algorithms we developed. We validate the usefulness of
the toolkit with three canonical applications that use the toolkit to generate
explanations for end-users.
Bio-data
Brian Lim is a 3rd year Ph.D. student at
the Human-Computer Interaction Institute in
Seminar 18
Title: User Experience Design in a MNC
Speaker: Dr Steven John Kerr
Chaired by: Jamie Ng Suat Ling
Venue:
Time: 16:00-17:00, November 8, Monday, 2010
Abstract
User experience design is a term some can
find confusing especially when other terms such as user centered design,
interaction design, information architecture, human computer interaction, human
factors engineering, usability, and user interface design are commonly used.
There are many overlaps, but these terms essentially refer to a toolbox of
processes and techniques, which can be applied depending on the situation and
available resources. They address the same goal of making things more simple
and enjoyable for users by making them the focus rather than the technology.
User Experience (UX for short) is associated heavily with the commercial sector
and is used as an encompassing term for the profession and professionals who
attempt to improve all aspects of the end users interaction with a product or
service. This presentation will discuss how UX is incorporated in a design
engineering environment for a large MNC and the different processes used in the
design and evaluation of customer products.
Bio-data
Before joining I2R, Steven
spent 3 years as the UX Lead for Software in Dell’s Experience Design Group in
Seminar 17
Title: Scenario-based Design: Methods & Applications in Domestic Robots
Speaker: Dr Xu Qianli
Chaired by: Tan Yiling Odelia
Venue:
Time: 14:00-15:00, October 14, Thursday, 2010
Abstract
Requirement development has been
considered as an important issue in product and systems design. It is of
particular interest to emerging products which market acceptance is untested
yet. It becomes essential to elicit user needs of these products from a design
perspective. However, existing requirement analysis methods such as card
sorting, focus groups, and ethnography study have not been able to include the
usage context in the design loop, thus jeopardizing the quality of the
developed requirements. In this regard, this research proposes a scenario-based
design approach to develop customer requirements. In line with the concept of
product ecosystems, scenarios are designed to facilitate the elicitation and
analysis of customer requirements. The scenario-based design approach includes
three major steps, namely, (1) scenario construction, (2) scenario deployment,
and (3) scenario evaluation. The proposed method is anticipated to tackle the
requirement development issues in the on-going project of robotic product
design. It is expected that the scenario-based approach will be useful to
include the customers experience in the requirement development, and enhance
the product development cycle so as to attain innovative solutions.
Bio-data
Before joining the CGI Department, I2R,
Qianli was a research fellow in the School of
Mechanical & Aerospace Engineering,
Seminar 16
Title: Designing for diversity - Exploring the challenges and opportunities of working with older adults
Speaker: Dr Mark David Rice
Chaired by: Jamie Ng Suat Ling
Venue:
Time: 14:30-15:30, September 3, Friday, 2010
Abstract
It is well documented that older people
provide much greater challenges to user-centered design than more traditional
mainstream groups. These can be related
to a wealth of different reasons – e.g. sensory loss, culture, language and
differing attitudes and experience to technology. Similarly, there is a great
necessity to understand how technology can support a rapidly ageing society,
requiring user-centered innovation by both private and public sectors to
develop new products and services for older people. Nevertheless, recent
research indicates that the ageing market lacks a growing need for human
dimension and a focus on individuals and relationships rather than
technological systems per se. During this talk I will give an overview of the
most significant aspects of my work in human-computer interaction with older
people. Amongst the various examples given, I will present some results of a
three year European study utilizing the digital television platform to deliver
a ‘brain training’ program for older adults. This includes presenting some of
the practical issues of working with an extensive panel of volunteers (+ 300),
the user-centered design methods applied and how they fitted within the
technical development work undertaken, as well as highlighting the business and
exploitation plan for the potential target market.
Bio-data
Mark joined the Human Factors Engineering
group in CGI this August. Previous to moving to
Seminar 15
Title: Product Ecosystem Design: Affective-Cognitive Modeling and Decision-making
Speaker: Dr Xu Qianli
Chaired by: Jamie Ng Suat Ling
Venue:
Time: 10:30-11:30, June 17, Thursday, 2010
Abstract
Design innovation that accommodates
customers’ affective needs along with their cognitive processes is of primary
importance for a product ecosystem to convey more added-value than individual
products. Product ecosystem design involves sophisticated interactions among
human users, multiple products, and the ambience, and thus necessitates
systematic modelling of ambiguous affective states in
conjunction with the cognitive process. This research formulates the
architecture of product ecosystems with a particular focus on activity-based
user experience. An ambient intelligence environment is proposed to elicit user
needs, followed by fuzzy association rule mining techniques for knowledge
discovery. A modular colored fuzzy Petri net (MCFPN) model is developed to
capture the causal relations embedded in users’ affective perception and
cognitive processes. The method is illustrated through the design of a product
ecosystem of a subway station. Initial findings and simulation results indicate
that the product ecosystem perspective is an innovative step toward
affective-cognitive engineering and the MCFPN formulism excels in incorporating
user experience into the product ecosystem design process.
Bio-data
Before joining the CGI Department, I2R,
Qianli was a research fellow in the School of
Mechanical & Aerospace Engineering,
Seminar 14
Title: Category Level Object Detection and Image Classification using Contour and Region Based Features
Speaker: Alex Yong-Sang Chia
Chaired by: Huang Zhiyong
Venue:
Time: 10:30-11:30, May 13, Thursday, 2010
Abstract
Automatic recognition of object categories
from complex real world images is an exciting problem in computer vision. While
humans can perform recognition tasks effortlessly and proficiently, replicating
this recognition ability of humans in machines is still an incredibly difficult
problem. On the other hand, successful automatic recognition technology will
have significant and mostly positive impacts in a plethora of important
application domains like image retrieval, visual surveillance and automotive
safety systems. In this talk, we present our novel contributions towards two
main goals of recognition: image classification and category level object
detection. Image classification seeks to separate images which contain an
object category from other images, where the focus is on identifying the
presence or absence of an object category in an image. Object detection
concerns the identification and localization of object instances of a category
across scale and space in an image, where the goal is to localize all instances
of that category from an image. We will present our method which exploits
contour only features for recognition. This method has achieved robust
detection accuracy, where it obtained the best contour based detection results
published so far for the challenging Weizmann horse dataset. Additionally, we
will also present a flexible recognition framework which fuse contour with
region based features. We show that by exploiting these complementary feature
types, better recognition results can be obtained.
Bio-data
Alex Yong-Sang Chia
received the Bachelor of Engineering degree in computer engineering with
first-class honours from Nanyang
Technological University (NTU),
He was awarded the Tan Sri Dr. Tan Chin Tuan
Scholarship during his undergraduate studies. In 2006, he received the A*STAR
Graduate Scholarship to pursue the Ph.D. degree in the field of computer
vision. He was also awarded the Tan Kah Kee Young Inventors' Merit Award in the Open Category in
2009 for his contribution towards object detection algorithms.
Alex Chia's research
interests are focused on computer vision for object recognition and abnormality
detection, feature matching under large viewpoint changes, and machine learning
techniques.
Seminar 13
Title: Supporting Intelligibility in Context-Aware
Applications
Speaker: Brian Lim, Human-Computer Interaction Institute, CMU
Chaired by: Jamie Ng Suat Ling
Venue: Turing at 13 South, Connexis,
Fusionopolis
Time: 10:30-11:30, December 23, Wednesday, 2009
Abstract
Context-aware applications employ implicit
inputs, and make decisions based on complex rules and machine learning models
that are rarely clear to users. Such lack of system intelligibility can lead to
the loss of user trust, satisfaction and acceptance of these systems.
Fortunately, automatically providing explanations about a system’s decision
process can help mitigate this problem. However, users may not be interested in
all the information and explanations that the applications can produce. Brian
will be presenting his work on what types of explanations users are interested
in when using context-aware applications. This work was conducted as an online
user study with over 800 participants. He will discuss why users demand certain
types of information, and provide design implications on how to provide
different intelligibility types to make context-aware applications intelligible
and acceptable to users.
Bio-data
Brian Lim is a 3rd year Ph.D. student at
the Human-Computer Interaction Institute in
Website: www.cs.cmu.edu/~byl
Seminar 12
Title: Automatic and Real-time 3D face reconstructions
Speaker: Dr Nguyen Hong Thai, CGI, I2R
Chaired by: Chin Ching Ling
Venue: Potential 2 at 13 North, Connexis,
Fusionopolis
Time: 16:30-17:00, December 9, Wednesday, 2009
Abstract
Automatic generation of realistic 3D human
face has been a challenging task in computer vision and computer graphics for
many years with numerous research papers published in this field attempting to
address this problem. With growing number of social networks and online-game,
the need of fast, good and simple system to reconstruct 3D face is rising. This paper describes a system for automatic
and real-time 3D photo-realistic face synthesis from a single frontal face
image. This system employs a generic 3D
head model approach for 3D face synthesis which can generate the 3D mapped face
in real-time.
Bio-data
Hong Thai received his Ph.D. in Electrical
Engineering and Information Technology from
Seminar 11
Title: Are Working Adults Ready to Accept e-Health at Home?
Speaker: Tan Yiling, Odelia, CGI, I2R
Chaired by: Chin Ching Ling
Venue: Potential 2 at 13 North, Connexis,
Fusionopolis
Time: 16:00-16:30, December 9, Wednesday, 2009
Abstract
Many researchers and designers from the
private and government sectors have been paying much attention to the issues,
policies, challenges and opportunities of bringing healthcare services to
patients’ homes. Some believe that this would not only ease the workload of the
healthcare practitioners but also reduce the cost of healthcare treatments.
Many topics surround the older adults but few focus on current working adults,
who may be the users of the future e-Health systems and policies. Compared to
older adults, the younger working adults are more tech-savvy and hence more
likely to adopt and afford technology-mediated health services. This paper
discusses the view points of working adults concerning the adoption of e-Health
as well as addresses issues concerning innovation and acceptance in their
future homes.
Bio-data
Odelia received her Masters of
Science in Technopreneurship and Innovation from NTU
in 2008. She joined the Institute for Infocomm Research,
A-STAR, as a Research Officer, and is currently working in the Home2015 (Phase
1) and P3DES Programme. Her research interests
include human factors in smart homes, healthcare and entertainment and human
computer interaction.
Seminar 10
Title: Control of Nonlinear Systems with
Speaker: Dr Tee Keng Peng, CGI, I2R
Chaired by: Dr Huang Zhiyong
Venue: Turing at 13 South, Connexis,
Fusionopolis
Time: 16:00-16:30, November 24, Tuesday, 2009
Abstract
This paper presents a control for
state-constrained nonlinear systems in strict feedback form to achieve output
tracking. To prevent states from violating the constraints, we employ a Barrier
Lyapunov Function, which grows to infinity whenever
its arguments approaches some limits. By ensuring boundedness
of the Barrier Lyapunov Function in the closed loop,
we guarantee that the limits are not transgressed. We show that asymptotic
output tracking is achieved without violation of state constraints, and that
all closed loop signals are bounded, provided that some feasibility conditions
on the initial states and control parameters are satisfied. Sufficient
conditions to ensure feasibility are provided, and they can be checked offline
by solving a static constrained optimization problem. The performance of the
proposed control is illustrated through a simulation example.
Bio-data
Keng Peng received his PhD in
Control Engineering from NUS in 2008. He joined the Institute for Infocomm Research, A-STAR, as a Research Engineer, and is
currently working in the Inter-RI Robotics Programme.
His research interests include adaptive nonlinear control, robotics, and human
motor learning.
Seminar 9
Title: Development of a computational cognitive architecture for intelligent virtual character
Speaker: Dr Liew Pak-San, CGI, I2R
Chaired by: Dr Huang Zhiyong
Venue: Turing at 13 South, Connexis,
Fusionopolis
Time: 14:30-15:00, October 12, Monday, 2009
Abstract
A development of a computational cognitive
architecture for the simulation of intelligent virtual characters is described
in this paper. By specializing and adapting from an existing structure for a
situated design agent, we propose three process models—reflexive, reactive and
reflective—which derive behavioural models that
underlie intelligent behaviours for these characters.
Various combinations of these process models allow intelligent virtual
characters to reason in a reflexive, reactive and/or reflective manner
according to the retrieval, modification and reconstruction of their memory
contents. This paper offers an infrastructure for combining simple reasoning
models, found in crowd simulations, and highly deliberative processing models
or reasoning, found in ‘heavy’ agents with high-level cognitive abilities.
Intelligent virtual characters simulated via this adapted architecture can
exhibit system level intelligence across a broad range of relevant tasks.
Bio-data
Pak-San received his PhD in Design
Computation from the Key Centre of Design Computing and Cognition,
Seminar 8
Title: Tennis Space: An interactive and immersive
environment for tennis simulation
Speaker: Dr
Chaired by: Dr Huang Zhiyong
Venue: Turing at 13 South, Connexis,
Fusionopolis
Time: 14:00-14:30, October 12, Monday, 2009
Abstract
This talk reports the design and
implementation of an interactive and immersive environment (IIE) for tennis
simulation. The presented design layout, named Tennis Space, provides the
necessary immersive experience without overly restricting the player. To
address the instability problem of real-time tracking of fast moving objects, a
hybrid tracking solution integrating optical tracking and ultrasound-inertial
tracking technologies is proposed. An L-shaped IIE has been implemented for
tennis simulation and has received positive feedback from users.
Bio-data
Shuhong is currently working on the Interactive Sports Game
Engine (ISGE) project. Before joining the department of computer graphics &
interface, I2R, he worked at
Seminar 7
Title: Real time tracking technologies for Augmented Reality
Speaker:
Chaired by: Dr Huang Zhiyong
Venue: Turing at 13 South, Connexis,
Fusionopolis
Time: 14:00-14:30, September 18, Friday, 2009
Abstract
Augmented Reality generally refers to the
splicing of the virtual world onto the real one, so that the user perceives
both at the same time. The focus of this talk is on attaching 3D graphics onto
real moving objects. This requires accurate tracking of the position and
orientation of the objects with respect to the viewer or camera. Tracking
systems based on computer vision, inertial sensing and differential GPS will be
presented. In particular, gyroscope triad calibration without
external equipment, a novel differential GPS technique, and accurate computer
vision tracking of planar objects without markers, which is robust to
illumination interferences and partial occlusions, are highlighted.
Bio-data
Louis Fong received his B.Comp from the
Seminar 6
Title: Computer Photomatic Stereo
and Weather Estimation with Internet Image
Speaker: Dr
Chaired by: Dr Huang Zhiyong
Venue: Turing at 13 South, Connexis,
Fusionopolis
Time: 10-10:30, August 25, Tuesday, 2009
Abstract
With the growth and maturity of search
engine, there is a trend to exploit how to make use of internet data to tackle
difficult classical vision problems, and enhance the performances of current
vision techniques. This talk will present a technique which extends photometric
stereo to make it work with much diversified internet images. For popular
tourism sites, thousands of images can be obtained from internet search
engines. With these images, our method computes the global illumination for
each image as well as the scene geometry information. The weather conditions of
the photos can then be estimated from the illumination information by using our
lighting model of sky.
Bio-data
Seminar 5
Title: Computer vision-based interaction and registration
for augmented reality systems
Speaker: Dr Yuan Miaolong, CGI, I2R
Chaired by: Dr Huang Zhiyong
Venue: Three Star Theatrette
Time: 11-11:30, May 30, Friday, 2008
Abstract
Augmented reality (AR) is a novel
human-machine interaction that overlays virtual computer-generated information
on a real world environment. It has found good potential applications in many
fields, such as military training, surgery, entertainment, maintenance and
manufacturing operations. Registration and interaction are two key issues which
currently limit AR applications in many areas. In this talk, I will introduce
some vision-based interaction tools and registration methods which have been
integrated into our augmented reality system. Some videos will be shown to
demonstrate the related methods.
Bio-data
Yuan Miaolong
received his BS degree in Mathematics from 
interactive tools.
Seminar 4
Title: A model of human motor adaptation to stable and unstable interactions
Speaker: Dr Tee Keng Peng, CGI, I2R
Chaired by: Dr Huang Zhiyong
Venue: Three Star Theatrette
Time: 11:30-12, May 23, Friday, 2008
Abstract
Humans have striking capabilities to perform
many complex motor tasks such as carving and manipulating objects. This means
that humans can learn to compensate skillfully for the forces arising from the
interaction with the environment. As an attempt to understand motor adaptation,
this work introduces a model of neural adaptation to novel dynamics and
simulates its behavior in representative stable and unstable environments. The
proposed adaptation mechanism, realized in muscle space, utilizes the stretch
reflex to update the feedforward motor command, and
selective deactivation to decrease coactivation of
agonist-antagonist muscles not required to stabilize movement. Simulations on a
2-link 6-muscle model show that motion trajectories, evolution of muscle
activity, and final endpoint impedance are consistent with experimental
results. Such computational models, using only 
measurable variables and simple computation, may be
used to simulate the effect of neuro-muscular
disorders on movement control, to develop better controllers for haptic devices and neural prostheses, as well as to design
novel rehabilitation approaches.
Bio-data
Tee Keng Peng received the B.Eng degree
and the M.Eng degree from the National University of
Singapore, in 2001 and 2003 respectively, both in mechanical engineering. Since
2004, he has been pursuing the Ph.D. degree at the Department of Electrical and
Computer Engineering, National University of Singapore. In 2008, he joined the
Neural Signal Processing Group at I2R as a research engineer. His
current research interests include adaptive control theory and applications,
robotics, motor control, and brain-computer interfaces.
Seminar 3
Title: Personalising a Talking Head
Speaker: Dr Arthur Niswar,
CGI/P3DES, I2R
Chaired by: Dr Huang Zhiyong
Venue: Three Star Theatrette
Time: 11:30-12, May 9, Friday, 2008
Abstract
Talking head is a facial animation system
which is combined with a TTS (Text-To-Speech) system to produce audio-visual
speech. The facial animation system is built by creating the head/face model with
the required parameters for speech animation. To build the head model, the
facial data of the subject have to be recorded. For this, usually hundreds of
markers have to be put on the subject's face, which is a laborious process,
especially if one wants to create the head model for another person. This
process can be simplified by modifying the previously constructed head model
using two images of the person, which is the subject of this talk.
Bio-data
Arthur Niswar
got his B.Eng. in Electrical Engineering from Bandung Institute of Technology (
Seminar 2
Title: How Creating Pervasive and Smart Projected Displays
Speaker: Dr Song Peng, CGI/P3DES, I2R
Chaired by: Dr Huang Zhiyong
Venue: Three Star Theatrette
Time: 11:30-12, May 2, Friday, 2008
Abstract
Projectors and cameras nowadays are
becoming less expensive, more compact and mobile. With off-the-shelf equipment and
software support, projected displays can be set up quickly anywhere on any
surfaces, breaking the traditional confines of space limits and projection
surfaces. In order to create desirable displays using projectors and cameras,
there are issues to be addressed, such as geometric and photometric
distortions, out-of-focus blurring, etc. In this talk, an overview of the
projector-camera systems will be introduced, followed by the problems and
solutions in creating pervasive and smart projected displays.
Bio-data
Song Peng
received his B.S. in Computer Science and B.A. in English Language from
Seminar 1
Title: How is image information processed in the human visual system (HVS)
Speaker: Dr Tang Huajin, CGI/P3DES,
I2R
Chaired by: Dr Huang Zhiyong
Venue: Three Star Theatrette
Time: 11:30-12, April 18, Friday, 2008
Abstract
This talk will cover some important
discoveries so far on the structure and functions of human visual system. The
core area in the brain, the primary visual cortex is believed to organize to
realize some fundament functions, such as edge detection and motion detection
and then to fulfill the high level perception including object recognition. The
experimental procedure to investigate the visual cortex is introduced, and the
computational model that underlies the computational principles of visual
system is also presented. The aim of the research work is to establish a bridge
between neuroscience and computer science in the area of visual perception, and
to elicit novel, intelligent and efficient methods for image analysis, computer
vision, etc.
Bio-data
Huajin Tang received the B.Eng and
M.Eng degrees from
He has authored or coauthored a number of papers in
peer-reviewed international journals, including IEEE Trans. on Neural Networks,
Circuits & Systems, Neural Computation, Neurocomputing,
etc. He has also coauthored one monograph in 2007 published by Springer in his
research area. His research interests include machine learning, neural 
networks, computational and biological intelligence.