Title: Television in Words TIWO Round Table Project Launch Meeting EPSRC GRR6719401
1Television in Words TIWO Round TableProject
Launch Meeting EPSRC GR/R67194/01
- University of Surrey
- 17th December 2001
2Vision and Language the challenge for computer
science
- Intelligent Multimedia Information Retrieval
- Want intuitive access to multimedia archives
especially with advent of TV / Internet crossover - Machine-level view of image/video data is
low-level ? need to represent high-level
information about content - May use collateral text for image/video
annotation, or use a combination of media for
communication ? - ?How to integrate unstructured, non-symbolic
image and video data with structured, symbolic
text data in information systems?
3Vision and Language an interdisciplinary
approach
- Researchers in artificial intelligence borrow
theoretical constructs and methods from
philosophy, aesthetics, semiotics, psychology,
and linguistics to help in the development of
systems, data structures and algorithms for
intelligent systems - The correspondence problem
- How is collateral information related in an
intelligent system? And, how to convert from
one form to another? (e.g. via knowledge
representation formalisms or via neural networks)
4Vision and Language Work in Progress at Surrey
- Prof. Khurshid Ahmad
- Dr Andrew Salway, Dr Andrew Hippisley, Dr
Lilian Tang - RAs Lee Gillam, Bogdan Vrusias
- PhD Students Chris Handy, Mariam Tariq, David
Boulton, Yan Xu, Elia Tomadaki
5Research into Images and Texts Work in Progress
at Surrey
- Digital Dance Library
- Scene-of-crime Information System
- Virtual Art Gallery
- Automatically Annotating Medical Images
- Royal Society Archives
- Television in Words
6Research into Images and Texts Work in Progress
at Surrey
- Characterise a special language used to describe
and interpret visual information corpus
linguistics - Extract information from collateral text in order
to index image/video data natural language
processing and object-relational databases - Model relationship between image-text to provide
intuitive access to digital libraries containing
a mix of verbal and visual information data
modelling and hypermedia
7Digital Dance Library
8SOCIS Scene-of-crime Information System
- I can see what appears to be a male laying in the
prone position on the floor. - He is wearing a maroon striped shirt with white
collar and cuffs, blue jeans, and has a pair of
left and right training shoes which have become
slightly dis-extended from the foot. - There appears to be a green tie down by his right
hand and I can see a possible footwear impression
in blood on his right hand. - Surrounding the body there are droplets of blood,
footwear impression in blood and several pieces
of broken glass and bottles.
9Image and Text at TATE Britain
at the extreme right, appears a scandalously
hieratic-looking couple a young mother is
strolling with her little girl dressed in white
with a salmon-colored sash
10TIWO Aims
- Continuing research into the link between vision
and language for intelligent multimedia
information systems - Develop a computational framework for
investigating narrative in multimedia systems - Adapt and apply existing video data models and
knowledge representation schemes to the task of
video annotation with audio description scripts - Adapt and apply existing language processing
techniques to process audio description scripts
11TIWO approach
- Gather an audio description corpus
- Analyse language used for audio description
- Evaluate information extraction technology for
processing audio description scripts - Evaluate data models and knowledge representation
formalisms for video data - Specify, design and prototype AuDesc a system
to prepare and manage audio description scripts - ?? Style checking
- ?? Customisation of scripts
- ?? Process scripts into video annotations for
retrieval
12TIWO objectives
- A digital library comprising video material of
diverse programme types, audio description, and
machine-executable representations of video
content. - A specification and prototype of a system
(AuDesc) to assist in the preparation and
management of audio descriptions - Two papers at international conferences (e.g. ACM
Multimedia and AAAI in 2003/4) and two journal
submissions (e.g. AI, AI Review, IEEE Multimedia,
Multimedia Tools and Applications) - A PhD dissertation
- A major EPSRC research proposal to continue
research in this area
13TIWO workplan
- WP1 Adapt and Apply Video Data Modelling and
Knowledge Representation 6 p.m. (months 1-10) - WP2 Adapt and Apply Language Engineering
Techniques 9 person months (months 4-24) - WP3 Audio Description System 12 p.m. (months
7-30) - WP4 Audio Description Round Table 1 p.m.
(months 1-36) - WP5 Write up of PhD Dissertation 8 p.m.
(months 11-12 31-36) - 1 full-time PhD student Yan Xu
- 1 associated PhD student Elia Tomadaki
- 5 hours per week from AJS (Principal
Investigator) - Official Start Date 2 January 2002
14The role of the TIWO Round Table
- RNIB
- Denise Evans (Head of Broadcasting and Talking
Images) - BBC
- Jane Westrop (Audio Description Manager)
- ITFC
- Garry Duguid (Director of Engineering and
Technology) - Softel
- Russ Wood (Research Director)
15The role of the TIWO Round Table
- Data and Expert Knowledge
- Domain understanding and modelling
- Providing sample audio description scripts
- Eliciting audio description in controlled
scenarios
16The role of the TIWO Round Table
- Current Issues and Technology
- What current problems should we be addressing?
User requirements. - What is the state-of-the-art in audio description
technology?
17The role of the TIWO Round Table
- User-feedback
- We would be grateful for continuous feedback on
the prototype AuDesc system - demonstrations at Round Table meetings and
on-site visits - latest information on TIWO WWW-site
18The role of the TIWO Round Table
- Technology Transfer and Project Dissemination
- The ideas / technology developed and evaluated in
the project are made available to the Round Table
first - Please tell your colleagues about the project!
19Television in Words TIWO Round TableProject
Launch Meeting EPSRC GR/R67194/01
- University of Surrey
- 17th December 2001