 |
RESEARCH
 |
|
|
My research area is the interaction design for multimedia retrieval systems, particularly digital video retrieval systems.
The underlying signal processing and video indexing techniques are being researched within my Centre, including shot/scene
boundary detection, news story segmentation, keyframe extraction, person/object detection and tracking, video summarisation and
personalisation, searching and browsing techniques. These methods, while all potentially useful, need to be set in a proper
context and possible scenario, with well-designed user-interfaces for a user to interact with in a meaningful way.
My role in the Centre is to connect these underlying
development and various techniques into a feasible, usable front-end user interface, so that we could envisage a (near) future
where these system would be commonly used in some specific place for some specific task by some specific group of users.
I use scenario, sketching and prototyping to do this activity; I conduct user study with usability engineering methods to assess
and refine the systems we developed.
More recently, object-based operations are developed in the Centre, and thus I am working on developing
object-based interactions, to be able to allow the user to view and specify objects within the video content and use them to
further search the related objects.
Have a look at the
 Summary of Interfaces within the centre. These include
fully deployed systems with a large number of users, as well as in-lab systems.
|
| research centre |
Centre for Digital Video Processing
led by Prof. Alan F. Smeaton
The Centre is developing various automatic content-based techniques for video, its approach is to develop concrete and complete
systems that feature these techniques so that we can draw the future usage of such systems.
A series of digital video demonstrator systems called Físchlár,
which allow users to browse & search digital video archives, is a good example.
Latest application of the Físchlár family is called Físchlár-News,
which archives daily TV news, automatically indexes by various content-based operations including shot/story boundary detection,
face detection, searching, threading, and personalisation. The system contains more than 7,600 linked news stories
all searchable, browseable, and playable on a conventional web browser. I have conducted a longitudinal user study of
this system, involving monitoring 16 users for a
1-month period during which I collected interaction log data and incident diaries
(see my ACM TOIS article 24(2), for full detail).
|
| on video interfaces |
My work involves, within the domain of multimedia IR and especially digital video IR:
- Creating novel interaction schemes
- Designing concrete (widget-level) user-interfaces
- Conducting user study with the designed interfaces
- Refining interfaces based on user study findings
Starting in 1998 with initial design of Físchlár, I have been designing a number of experimental user-interfaces which
have been mounted on top of various technical systems with automatic content-based operations and functionality that our centre
develops. Target design platforms include the Web, mobile, TableTop, and interactive TV.
|
|
|
|
