Virtualized Classroom – Automated Production, Media Integration

and User-Customized Presentation

Zhigang Zhu1,2*, Chad McKittrick1 and Weihong Li2
1
Department of Computer Science, City College, The City University of New York, New York, NY 10031
2
Department of Computer Science, Graduate Center, The City University of New York, NY 10016
*zhu@[Link]

Abstract students) in the classroom is collected and

Multimedia materials from classrooms and analyzed, using multiple sensors (cameras,
seminars are rich sources of information. Our microphones, screen capturer and whiteboard
Virtualized Classroom Project, which is a testbed for digitizer). Automatic camera management is
integrating novel multimedia software systems, performed in this stage with emphasis on three
addresses fundamental research problems occurring in major tasks: automated camera control, media
different components of a next-generation, processing- synchronization and automatic data collection.
enhanced e-learning system. Our ultimate goal is to Intelligent post-processing: compression, cross-
create a natural learning environment that could give media indexing and archiving. We will develop
students interactive experiences equivalent to, or even techniques to perform post-processing for the cross-
better than, the real classroom. This will be enabled indexing of materials via an indexed Table of
by virtual instructor presence, user-selectable and Contents of keywords and topics, and for the
fully-indexed visual aids, and flexible communications purposes of subsequent information search and
between students, the instructor and virtual instructors retrieval (e.g., using cross-indexed text, audio, and
(computers). We have developed several basic image to enhance retrieval using any one of these
components for the Virtualized Classroom project: modalities), subsequent media substitution (e.g.,
automated data collection, intelligent media user replacement of PowerPoint images in video
integration, and flexible user interfaces. with digital representations drawn directly from the
original PowerPoint slides), and subsequent user
customized lecture presentation.
1. Introduction Natural Interaction between students, teachers
A domain of primary importance in the future of and computers. One of the main concerns of
web-based technology and digital libraries is distance today’s e-learning systems is whether deep
and electronic education (e-learning). Multimedia understanding of concepts(not just the
materials from classrooms and seminars are rich accumulation of facts) could occur in the absence of
sources of information. Today's first generation e- human interaction. In the Virtualized Classroom,
learning systems primarily adopt a “record-and- we are aiming to create a natural environment that
playback” approach, which does not leverage the could improve the real classroom in stimulating
processing capabilities (during live capture, after- broader and novel interaction, -- enabled by the
capture processing, and during later user interaction concept of active media objects in the Virtualized
with archived materials) that we believe will underlie Classroom, including: a virtual instructor presence,
the next generation of more automated, flexible, user-selectable and fully-indexed visual aids and
comprehensible, and individualized e-learning systems. materials, full communications between students,
The goal of the Virtualized Classroom project is to instructors and computers, all organized in the
design novel software systems to address fundamental Virtualized Classroom environment that is natural
research problems occurring in different components and friendly.
of a next-generation, processing-enhanced e-learning We use the name Virtualized Classroom to indicate
system for college and graduate courses. It includes that the new e-learning environment is generated and
the following key components: enhanced from materials captured from real
Automatic data collection, analysis, and multi- classrooms. The Virtualized Classroom concept could
modal synchronization. In the pre-production be viewed as a special application of the Virtualized
stage, the audio/video and content information (of Reality technique [Kanade99]. We envision that every
blackboard/ whiteboard, slide projections, digital meaningful object is active in the Virtualized
slides, and activities of the instructor as well as Classroom.
 in the space (Fig. 1). Finally we are going to provide an
active question -answering window so that students
can share their discussions with each other and the
instructor.
The combination of active objects with the user-
(a) (b) (c)
customized interface provides effective interaction
between the teacher and students, and exactly follows
Active WWW link
the principles of good teaching: for example it
Synthetic digital display
encourages student to faculty and student to student
Slide projection in original
contact and active learning, it provides prompt
Panoramic mosaic from
feedback, and, it respects diverse ways of learning.
With very cost-effective multimedia sensors, we have
developed the following basic components for the
Virtualized Classroom project - automated data
collection, intelligent media processing and integration
Annotated
algorithms, and user-customized interface designs. We
instructor will describe each of the components in the following
three sections. Related work will be discussed in
(d) Section 5. Then we will discuss some of the future
research directions in the Virtualized Classroom.
Finally we will conclude our work in Section 7.
Fig. 1. Active objects and interactive user
interfaces. A mosaic is generated from the panning
camera tracking the instructor [Zhu99]. (a) Frame 1, 2. Automated Data Collection
(b) Frame 50 and (c) Frame 150. (d) Improved
panoramic mosaic with the slide projection aligned The commonly used classroom /lecture presentation
with the high resolution digital slide. Note that tools are PowerPoint (PPT) slides, overhead
active objects of the digital display enable active projections, and blackboards /whiteboards. In order to
WWW links in the images that the user can click on. digitize the classroom/lecture contents into computers,
we use a low-cost Mimio digital whiteboards system to
First the image of the instructor is active not only substitute the use of the blackboard. The "sensors" we
because it is dynamic but it also is annotated and are using for the classroom setting are PPT slides
linked to his/her important information such as his/her capturer, Mimio Virtual Ink and Sony video camera
web page, email and phone numbers. By clicking the with a RemoteReality omnidirectional lens. All the
image of the instructor, a student can find this sensors can be easily managed by the
information and may also ask questions on the spot, or instructor/lecturer himself/herself.
make an appointment by sending him/her email, or
sometimes even directly talk with him/her (if he/she is
online). In addition, students may be able to engage in
discussion with the instructor and other students about
key points in the lecture through a discussion board in
the Virtualized Classroom interface. Second the
synthetic images of the digital slides and blackboard
are also active with a glossary and dictionary, remedial
tutorials, WWW links, and audio/video clips that may
be presented in the original lecture. A Table of
Contents (TOC) is active and at hand in the sense that
it is floating or hiding in the 3D virtual space so that it Fig. 2. Powerpoint © slide capture. The pop-up
can be activated whenever desired. The idea of active dialog-box of the PPT slide capturer add-in is
slides make it possible for a student to point to a term activated automatically when the instructor starts
or sentence that he/she wants to learn more about it, his/her PPT presentation.
and the system will search the notes in this lecture and 2.1. PowerPoint slide capture
across other lectures and designated courses, as well as The PowerPoint slide capturer was modified from
the collateral materials on the web as specified by the the Berkeley PPT Recording Add-In [BMRC, PINY-
instructor. The active slides are similar to HTML ELN]. We added the start date and time information of
pages, but they are presented in the virtual space the presentation under recording in order to
aligned with the instructor and other meaningful items synchronize the recorded PPT slides with the
accompanying whiteboard pages (Fig. 2). In addition to the Mimio Virtual Ink system. The instructor's video
the pages of slides in one of the image formats (e.g. and audio streams will be captured by an
JPEG), a PPT log file is automatically generated with omnidirectional camcorder. However, our Virtualized
timing information and the titles of all the slides in the Classroom authoring tool (creator) and presentation
presentation. tool (player) also work with a PPT-slides-only
presentation, a whiteboard-only presentation, an
[Link] capture audio/video-only presentation, or a combination of any
Everything an instructor writes on a normal whiteboard of them. After the instructor sets up the sensors and
can be captured by the Mimio Virtual Ink [MVI], a starts the presentation, almost everything is
hardware and software package for digitizing automatically saved for him/her.
whiteboard handwriting and drawings (Fig. 3). The
whiteboard presentation will be saved in a series of
html files, one html file (with the time information of
the page) for each whiteboard page (as JPEG image).
An [Link] file is also generated, which will be
used to retrieve the whiteboard pages. The Virtual Ink
system can also be used as a remote mouse and/or a
remote keyboard, in that the lecturer could control
his/her PPT presentation or even generate the PPT
presentation just in front of the whiteboard, without
touching the computer after he/she starts the system.
Fig. 4. The omnidirectional camera and an
omnidirectional image with both the instructor and
the students in view.

3. Media Processing and Integration

In this section, we will briefly describe several
computer vision techniques in media processing and
integration including panoramic viewing, media
Fig. 3. Mimi© virtual ink system is capable of
recording handwriting on a normal whiteboard of
synchronization, and media integration. These
2.4 m x 1.2 m, with 100 dpi resolution. techniques will be the basis for further cross-media
indexing and user-controllable viewing of e-lectures.
2.3. Video / Audio Capture
We use an omnidirectional lens developed by 3.1. Panoramic video processing
RemoteReality [RR] with a Sony DV camcorder to The omnidirectional camera is used to capture in
capture the entire classroom in 360 degrees so that both real-time the 360-degree FOV classroom scene. In the
the instructor and students will be in the field of view video stream, both the instructor and the students can
(Fig. 4, Fig. 5). Of course the lecturer could just setup be seen. Currently we only transform an original
the DV camera to get a normal video capture. The circular video stream into a cylindrical panoramic
lecturer starts recording right when he/she clicks the video stream for better viewing (Fig. 5). In the future,
“OK” button in the PPT Recording Add-In pop-up we will study the best way to present the classroom
panel so that the video stream will be synchronized setting using omnidirectional vision technology, for
with the slides. In our current implementation, the example, instructor tracking, student tracking (for
video stream should be saved either in an .avi or in question-answering and other interaction), and
.mpg file. integration of video with slide/ whiteboard
If the lecturer only needs to save the audio stream, presentations.
he/she just needs to set up the computer microphone, 3.2. Media synchronization
and in the PPT recording Add-In, to simply check the The synchronization of the PPT slides, whiteboard
“Audio” check box. A .wav file will be saved, and it is pages and audio/video streams is enabled by a simple
synchronized with the slides. stream synchronization algorithm which uses the
For the best use of the above sensors, we assume timing information in the PPT log file, whiteboard log
that the instructor will use a computer projector to files, and the video/audio stream. There are three issues
project PPT slides on a whiteboard in his/her class. The in synchronizing PPT and whiteboard page
handwriting contents written on the whiteboard are on presentation.
top of the slides projections, and will be captured by
Fig. 5. Panoramic view of the classroom video. The instructor, slide projection, blackboard and students are
all in the field of view.

First, each PPT slide has a start time in the low-cost video camera that views both the slide
presentation, while each whiteboard page has a time projection and the whiteboard [Li03].
indicating when the page is done. Second, the number
of whiteboard pages will most probably be different Fig. 6. Slide and
from the number of the PPT slides. The instructor may whiteboard
generate more than one white page within the image
presentation of a single slide, or he/she may use the registration.
Four crosses
same whiteboard page for several PPT slides. In other shown in (a) are
occasions, he/she may not generate any whiteboard the four corners
pages for some slides. Third, the start times of the (a) of the projection
corresponding PPT slides and the whiteboard pages region of PPT
may be different. Our algorithm matches up the two slides. Both (a)
timing sequences, taking care of all these issues. and (b) show the
size of the
3.3. Registering slides and whiteboard images Mimio
The PPT slide presentation and the whiteboard whiteboard
system when used together could provide a better active area
means to present a lecture in a classroom. When the (within the
(b) picture frames).
PPT projection and the whiteboard are in two separate
areas, we may not need to register PPT slide images
and the whiteboard images in the Virtualized 3.4. Registering slides and instructor images
Classroom presentation. However when the PPT slides The position and gestures of the instructor are a
are projected on the whiteboard, where the instructor is very effective way to attract the attention of students
going to annotate, modify, or expend the PPT and help them to recall what they have learned in the
presentation, we need to geometrically register the class. In the Virtualized Classroom, actual video
images from the two sources to create high quality images of the lecturer will be merged in real time with
presentations. As a simple and quick approach, right the digital slides, so that a student will perceive the
after he/she starts the classroom presentation, we natural spatial relation between the lecturer and the
require the instructor to mark at least the four corners visual aids. The lecturer or his/her “avatar” is in the
of the PPT projection in the whiteboard as a Virtualized Classroom. For example, the lecturer may
“calibration” step (Fig. 6a). The whiteboard image with point to the items in the digital slides (Fig. 7). For this
the four markers is captured by the Mimio Virtual Ink purpose, we have also done some preliminary research
and a corresponding PPT slide image is captured by on automatic instructor extraction from the video
PPT recorder. Since both of them are images of the streams using computer vision techniques in [Zhu00]
same plane (the whiteboard), we use a projective to perform content-based video compression and slide-
transformation to register the two images with (at least) video integration. As a result, the text always will be
four point pairs. Typically, the sources will remain presented in sharp resolution (using the digital slides),
stationary (e.g., the slide projector, whiteboards) and much higher than the standard type of video. The
we can use the same transformation parameters after image of the lecturer will be merged with the
we perform an initial registration. (Fig. 6b). For synthesized “slides” rather than displayed in two
dealing with more difficult cases where portable different windows.
presentation devices may be moved during a Fig. 7 shows an example of automatic integration of
presentation, we have developed algorithms that the real and high-resolution virtual images. In (a) is a
perform dynamic registration of the two sources via a
frame from the original video at low resolution with For Slides Panel, load the PPT log file (.txt)
undesired shadows. In (b) the extracted real image of For Whiteboard Panel, load the whiteboard
the lecturer and the synthetic image of the high [Link] file.
resolution digital slide are merged in one view, giving For Video with Audio or Video Panel, load the
the user a strong feeling that the lecturer is in the mpg or avi video file (up to two video streams can
Virtualized Classroom. The slide image from video be played)
and the synthetic image from digital slide have been For Audio file, load the .wav file
aligned so that the lecturer can point to the exact point For Slides with Board, load the PPT log file (.txt).
that he pointed in the slide projection. In (c) the image The VC Creator/Player assumes that the whiteboard
of the lecturer is displayed as a “shadow” on the slide information is in [Link] file, and the combined
projection to avoid occluding the slide, while still pages (from PPT and whiteboard images) have
providing a natural presence. In (d) the image of the been generated.
lecturer can be replaced by a colorful contour or even
simply make the item pointed to highlighted. The
highlighting requires gesture recognition and hand
localization.

(a) (b)
(a)

(c) (d)
Fig. 7. Slide and video registration
(b)
4. User-Customized Presentation Fig. 8. VCPS authoring tool interface. (a) Integrated
presentation of PPT and whiteboard pages; (b) PPT
Our current implementation of the Virtualized
slides + panoramic video
Classroom includes a Virtualized Classroom
Presentation System (VCPS) that is designed as both
After the user adds presentation components, the
an authoring tool and a presentation interface for
application window will contain frames encompassing
different kinds of lectures, and has a user-selectable
the presentation. The user can resize and reposition
interface. The VCPS that is developed in Java includes
any of the frames. Since the Creator works in a
two parts - the VCPS Creator and the VCPS Player.
WYSIWYG fashion, the frame window will be larger
4.1. Virtualized Classroom Creator than the component will be when it is played. So, even
The user (an instructor or a student) who uses the if frames overlap a bit, if the content in the frame is
system can customize the presentation by using the positioned as he/she wants, the presentation will play
VCPS Creator to include different media in windows as he/she wants. The two snapshots in Fig. 8 show two
with user-selected sizes and positions. After the user examples with different contents, number of windows,
opens a new Creator page, there will be a floating and sizes and positions of the windows.
frame that has checkboxes on it. He/She will use these Once the user has included all of the components in
checkboxes to add or remove presentation his/her presentation, he/she need to save the
components. Each time the user clicks on one of the presentation to a .vcp file. The operation is as simple
checkboxes a popup window will appear directing as to go to File menu item and to click on Save, and
him/her to click on a certain file so that the program then to chose a location in the file system, give the file
can load the proper information. The media forms that any name, and click on Save in the popup window.
have been integrated in the VCPS are the following:
4.2. Virtualized Classroom Player ability to support learner navigation, which is a key to
When the user wants to play a presentation all knowledge discovery and construction. Currently in
he/she has to do is click open under the Player menu. existing electronic lectures, text windows, video
It will take a few seconds for the presentation to load windows, and a table of contents (TOC) are typically
and to transform the images into the right sizes. Once in a fixed format.
it has loaded the user can click on the “Start” button to Active learning has been introduced in UMass
start the presentation from the beginning, or, if there RIPPLES group technologies by replacing the MANIC
are PowerPoint based slides, he/she can do a mouse brute force searching with more advanced searching
right click to get a slide based Table of Contents techniques [Kumela04] that make use of ranking and
(TOC) to guide the player to a specific topic in the relevance, and by applying an active object interface.
presentation. All the media contents are synchronized The new system generates the matching documents
by using the timing information (from the PPT log (e.g., relevant lecture segments) from within the locally
file, the whiteboard html file, and the video/audio available material. The content of the matching
timing information). Using a pop-up TOC, we can documents and the initial user query then are used to
make full use of the space in the player. Fig. 9 shows create a new query, conducted over the World Wide
snapshots of the VCPS Player interfaces. We will show Web and a database of reference material using
live demos for both authoring and playing VCPS e- Google. The returned links are filtered by relevance
lectures when presenting the paper. to learner interests. An “active” interface that generates
queries from student interactions and the capability to
query content of video and audio materials is also
desirable. Several commercial speech indexing
systems, including Streamsage [Sage] and Scansoft
[Scansoft], have been applied to indexing on-line
streamed courseware. The Informedia Digital Library
project at CMU [Cristel95, Wactlar99] is developing
new technologies for video analysis, storage, search,
and retrieval of image, video and audio content and
embedding them in a distributed library system for use
in education, training, sports and entertainment.
(a) Automated production is a major issue for e-lecture
systems because to improve the quality of
presentations many systems require significant manual
effort in analog data collection, digitization, and
synchronization. In the UMass RIPPLES program, as a
result of emphasizing low demand on instructors and
high quality presentations, it can take five hours labor
of a trained courseware developer to create a one-hour
lecture. We note that some systems have introduced
forms of automatic production. Cornell’s lecture
(b) browser [Mukho99] includes lecturer tracking, slide
Fig. 9. VCPS Player interface. (a) Integrated change detection and segmentation, and matching slide
presentation of PPT and whiteboard pages; (b) PPT
slides + panoramic video
projections with digital slides. A precursor to Cornell’s
lecture browser, UC Berkeley BMRC Lecture Browser
[BMRC] has several additional features including
5. Related Work audio search capabilities, bookmarks and an interactive
Projects focused on record and playback whiteboard for online students. To enable automated
technologies include: the MIT Physics Interactive production, we need a way to capture many classroom
Video Tutor project (PIVOT) [Lipson01], the Georgia activities. MIT Media Lab’s SmartCam [Pinhan95] has
Tech Classroom 2000 (eClass) [Abowd96, Abowd99], a robotic TV camera without a human operator,
the CMU Just-In-Time system [Dannen97, JITL], and changing its attitude, zoom and position to provide
the UMass Research in Presentation Production for specific images upon verbal request from a director.
Learning Electronically (RIPPLES, MANIC) The Intelligent Classroom [Frank99, Frank00] uses
[Stern97], among others. Most of these produce cameras and microphones to determine what the
courseware that is largely passive, typically only speaker is trying to do and then takes the actions (e.g.,
allowing some form of search over the text in the jumping to a new slide) it deems appropriate. Auto-
recorded “lectures.” They generally have a limited Auditorium [Cruz94, Brinchi98] uses multiple cameras
that automatically switch based on context. The the instructor, table of contents, and supplementary
automatic camera management capability (iCam) of materials into a spatial context.
Microsoft Research [Rui01, Liu01] automatically
tracks both the speaker and the audience in a lecture 6.3. Usability and case study
room environment by using omnidirectional-vision- We will study several typical classroom examples
and microphone-array-based techniques. The BlueEyes using different instructing technologies. For example,
project of IBM Almaden Research Center [IBM-BE] we will apply our VCPS presentation system to
uses non-obtrusive sensing technology, such as video different types of courses (science or art) at different
cameras and microphones, to identify and observe a levels (undergrad or graduate) and using different
user’s actions, and to extract key information, such as presentation methods (PPT presentation/ whiteboard/
where the user is looking and what the user is saying blackboard/speech-only). In different cases, we will
verbally and with gestures. study the data collection strategy and user interface
design. We will develop algorithms that can
automatically collect user feedback information for
6. Discussions and Future Work
assessing our Virtualized Classroom presentations.
In this section we discuss several on-going research
issues that enable fast presentation creation, active 6.4. Real-time presentation creation
navigation and natural interaction in the Virtualized The work to create a good PPT presentation is
Classroom. tedious, particularly when some fine drawings and
animations are generated. Therefore we will study the
6.1. Multimedia active object representations feasibility of using Mimio Virtual Ink or other
The ultimate goal of the project is to create e-lecture interactive tools to generate or to revise the PPT
presentations that include active objects in various presentation on the fly. For doing that, we need to
kinds of media - PPT slides, handwriting, and video design better interfaces than merely the available
and audio so that every meaningful object in the Mimio mouse function for interactive creation of
presentation is active - terms, pictures, instructor multimedia contents, and to develop algorithms for
images, and students. Collaborating with the UMass converting /combining handwriting text /drawings into
RIPPLES group, we intent to create a future PPT digital presentations.
Virtualized Classroom environment that supports
interactive learning even though the instructor and 7. Conclusions
students are neither in the same physical space
nor presented in the same time. Therefore content- In the Virtualized Classroom (VC) project, we have
based data collection and presentation is required. In tested very simple yet effective techniques to automate
the current version of our program, both the PPT slides the capturing of multimedia classroom materials,
and the whiteboard pages are saved as image files including audio/video, digital PPT slides and
(JPEG). This does not allow cross-indexing in and whiteboard writing. When we assume that the
among pages. In the next step, we are going to study instructor uses an LCD projector to project
ways to enable the implementation of the active object PowerPoint slides on a whiteboard in his/her class,
concept. For example we will save PPT slides into we use cost-effective sensors to collect multimedia
HTML files. We will save handwriting on whiteboards lecture materials in the classroom, and produce very
in their dynamic forms, and then further with effective e-lectures. Our automated PPT Slide Capturer
handwriting recognition. We will also extract the was modified from the Berkeley PPT Recording Add-
subjects (instructor and students) and the relevant In. Handwriting on the whiteboard (including notations
events (e.g. question & answering) from the video “over” projected images) is captured by the Mimio
sequence. Virtual Ink system. The instructor's video and audio
streams are captured by an omnidirectional camcorder
6.2. Interactive 3D user interface design (in order to see the full view of the classroom for
We will perform a feasibility study on the impact of students-teacher interaction). Our Virtualized
a 3D Virtualized Classroom environment on the active Classroom system can capture PPT-only presentations,
learning experience. With the 3D Virtualized whiteboard-only presentations, audio/video alone, or
Classroom, we allow users to gain better understanding any combination of these. Once the instructor sets up
of the 3D spatial context of the classroom environment. the sensors and starts the presentation, almost
We believe that this will directly improve student everything is automatically saved. The synchronization
perception of the material or classroom presentation. of slides, whiteboard pages and audio/video streams is
We will allow the user many choices for viewing the enabled by a stream synchronization algorithm that
lecture, organizing views of the blackboard, the slide, uses the logged timing information. A Table of
Contents (TOC) is automatically generated from the
log files. In the e-lecture presentation, the VCPS IEEE, Sept.-Oct. 1999. p.2-5.
system is designed as both an authoring tool and a [IBM-BE] IBM BlueEyes web page:
presentation interface for different kinds of lectures, [Link]
and implements a user-selectable interface. The [JITL] Just-In-Time Lectures,
automated data collection/synchronization techniques [Link]
and the user-customized interfaces are an important [Kanade99] T. Kanade, P. Rander, S. Vedula, and H. Saito,
part to the overall active learning environment in the Virtualized Reality: Digitizing a 3D Time-Varying Event
Virtualized Classroom. Our goal is to design a next As Is and in Real Time, Mixed Reality, Merging Real and
generation e-learning system that is cost-effective in Virtual Worlds, Y. Ohta, H. Tamura ed., Springer-
creating content, and low impact on instructors, while Verlag, 1999: 41-57
improving learning effectiveness (e.g., retention and [Kumela04] D. Kumela, K. Watts and W. R. Adrion,
student satisfaction) and presentation creation Supporting Constructivist Learning in a Multimedia
Presentation System, submitted to the Frontiers in
efficiency.
Education Conference, Savannah, GA October 2004
[Li03] W. Li, H. Tang and Z. Zhu, Automated registration of
8. Acknowledgements high-resolution images from slide presentations and
This work is partially supported by the CUNY whiteboard handwritings via a low-cost digital video
camera. The 2nd IEEE Workshop on Image and Video
Institute for Software Development and Design
Registration (with CVPR’04), July 2, 2004
(CISDD) and the CUNY Graduate Research
[Liu01] Q. Liu, Y. Rui, A. Gupta and J.J. Cadiz,
Technology Initiative (GRTI). We would like to thank Automating Camera Management in Lecture Room
Prof. Rick Adrion, Prof. Jim Kurose and Prof. Edward Environments , Proc. ACM CHI, Seattle, WA, 2001
Riseman at the University of Massachusetts at Amherst [MVI]. Mimio Virtual Ink,
for the fruitful discussions and collaboration in this [Link]
project. Our thanks are also given to Dr. Parviz [Mukho99] S. Mukhopadhyay and B. Smith, Passive
Kermani for sharing with us his experience and tools Capture and Structuring of Lectures, Proc. ACM
used in his e-learning lecture navigator (ELN) at the Multimedia, Orlando. FL, October 1999
Polytechnic University of New York. [Nayar98] V. Peri and S. Nayar, Generation of Perspective
and Panoramic Video from Omnidirectional Video, Proc.
9. References DARPA Image Understanding Workshop, 1997.
[Pinhan95] C. Pinhanez and A. Bobick, Intelligent studios:
[Abowd96] G. Abowd, et. al., Teaching and Learning as
using computer vision to control TV cameras, Proc.
Multimedia Authoring: The Classroom 2000 Project,”
IJCA'95 Workshop on Entertainment and AI/Alife,
Proc. ACM Multimedia ‘96, Boston MA, Nov 1996
Montreal, Canada, August, 1995.
[Abowd99] Gregory D. Abowd, Classroom 2000: An [PINY-ELN]. PINY ELN: e-Learning Lecture Navigator,
Experiment with the Instrumentation of a Living
[Link]
Educational Environment, IBM Systems Journal, Special
issue on Pervasive Computing, 38(4), 508-530, Oct. 1999 [Rui01] Y. Rui, L. He, A. Gupta and Q. Liú, Building an
Intelligent Camera Management System, Proc.
[Bianchi98] M. Bianchi, “AutoAuditorium: a Fully
Multimedia, Ottawa, Canada, Sept-Oct. 2001
Automatic Multi-Camera System to Televise Auditorium
Presentations, Proceedings of the Joint DARPA/NIST [RR]. RemoteReality, [Link]
Smart Spaces Workshop, Gaithersburg, MD, July 1998. [Sage] Streamsage, [Link]
[BMRC] BMRC Lecture Browser download at [Scansoft] Scansoft, [Link]
[Link] [Stern97] M. Stern, J. Steinberg, H.I. Lee, J. Padhye, J.
[Christel95] M. Christel, T. Kanade, M. Mauldin, R. Reddy, Kurose, MANIC: Multimedia Asynchronous Networked
M. Sirbu, S. Stevens, and H. Wactlar, Informedia Digital Individualized Courseware, Proc. Educational
Video Library. Communications of the ACM, 38(4), 1995 Multimedia and Hypermedia, 1997.
[Cruz94] G. Cruz and R. Hill, “Capturing and Playing [Wactlar99] H. Wactlar, et al, Lessons learned from the
Multimedia Events with STREAMS,” Proc. ACM creation and deployment of a terbyte digital video
Multimedia, San Francisco, CA, Oct. 1994 library, IEEE Computer 32(2): 66-73, 1999.
[Dannen97] R. B. Dannenberg, “Are Just-In-Time Lectures [Zhu00] Z. Zhu, K. D. Rajasekar, E. Riseman, A.
Effective At Teaching?” June 1997 Hanson. Panoramic Virtual Stereo Vision of Cooperative
[Link] Mobile Robots for localizing 3D Moving Objects. Proc.
IEEE Workshop on Omnidirectional Vision, Hilton
[Frank00] D. Franklin, S. Bradshaw, and K. Hammond.
Jabberwocky: You don’t have to be a rocket scientist to Head Island, June, 2000:29-36
change slides for a hydrogen combustion lecture. Proc. [Zhu99] Z. Zhu, G. Xu, E. M. Riseman, A. R. Hanson,
Int. Conf. on Intelligent User Interfaces, 2000. Fast Generation of Dynamic and Multi-Resolution 360
[Frank99] D. Franklin, J. Flachsbart, K. Hammond. The Panorama from Video Sequences, Proc. IEEE
Intelligent Classroom. IEEE Intelligent Systems, 14(5), ICMCS’99, Florence, June 7-11, 1999, vol 1, 400-406.