Transcript Viewing video clips

WP1- Contextual Feedback
WP1: Kieran Moran & Donal O’Gorman
WP1.1: Identify Perform. Factors Sarah Hughes
WP1.2: System Design & Implementation: RP3
WP1.3: Deployment & Data Gathering: Greg May
WP1.4: Content Analysis for Event Detection: Damien
Connaghan
WP1.5: Event Indexing for Retrieval: Colum Foley
WP1.6: Personalisation & Adaptation: Luke Conroy
UNIVERSITY COLLEGE DUBLIN  DUBLIN CITY UNIVERSITY  TYNDALL NATIONAL INSTITUTE
Summary of context info
For both players tag
ACTION TO/FROM Score
Serve
Forehand
Backhand
etc
Deep line
Short line
Service T
etc
Player
speed /
distance
15-0
OUTCOME
Win
Lose
Error
Rally
Ball Speed
Accuracy
Ball ht - net
PHYSIOL.
RESPONSE
Peak HR
Average HR
Sweat
etc
Allow for querying:
Statistics and Viewing video clips
[Link to Physiological response data]
UNIVERSITY COLLEGE DUBLIN  DUBLIN CITY UNIVERSITY  TYNDALL NATIONAL INSTITUTE
Allow for querying:
Statistics and Viewing video clips
[Link to Physiological response data]
Statistics:
•What % of serves from the right immediately
won rally
•What % of serves from the left resulted in a line
return that won rally (for the opposition)
•What % of drop shots won and lost rally
UNIVERSITY COLLEGE DUBLIN  DUBLIN CITY UNIVERSITY  TYNDALL NATIONAL INSTITUTE
Allow for querying:
Statistics and Viewing video clips
[Link to Physiological response data]
Viewing (It will be the coach/player who determines
why the outcome occurred – shot technique, shot
selection, movement speed, reading of the game
etc):
•Show all serves from the left that immediately won
the rally
•Show all the serves from the left the resulted in a
continued rally
•Show all the serves from the left that the server lost
within two strokes by the opposition
UNIVERSITY COLLEGE DUBLIN  DUBLIN CITY UNIVERSITY  TYNDALL NATIONAL INSTITUTE
This requires:
• Start, end and length of rally
• Location of shot (and initial position) [Player Tracking]
• Shot played
[Stroke recognition]
• Score
• (Ball speed)
UNIVERSITY COLLEGE DUBLIN  DUBLIN CITY UNIVERSITY  TYNDALL NATIONAL INSTITUTE
Which sensors will provide us with this info:
• Video (no intervention by coach/user)
• Ubisense (no intervention by coach/user)
• Ubisense/control unit (coach/user intervention)
[e.g. iTouch + 5 buttons, bluetooth microphone + ASR]
UNIVERSITY COLLEGE DUBLIN  DUBLIN CITY UNIVERSITY  TYNDALL NATIONAL INSTITUTE
Finally, we need to examine:
• How the coach should query the information
• How the information should be displayed
• What video view/information will be best to
show appropriateness of a player’s shot
technique, shot selection, movement speed,
reading of the game etc
• What is the best way of feeding this information
back to the player
UNIVERSITY COLLEGE DUBLIN  DUBLIN CITY UNIVERSITY  TYNDALL NATIONAL INSTITUTE
This requires:
• Start, end and length of rally
• Location of shot (and initial position) [Player Tracking]
• Shot played
[Stroke recognition]
• Score
• (Ball speed)
UNIVERSITY COLLEGE DUBLIN  DUBLIN CITY UNIVERSITY  TYNDALL NATIONAL INSTITUTE
Through data mining:
• Start, end and length of rally
• Location of shot (and initial position) [Player Tracking]
• Shot played
[Stroke recognition]
• Score
• (Ball speed)
UNIVERSITY COLLEGE DUBLIN  DUBLIN CITY UNIVERSITY  TYNDALL NATIONAL INSTITUTE
User interface design
• Designed a dummy user interface in Flex.
Features include: - the ability to draw
rectangles on a tennis court – this could be
used to data mine event 2.
• Developed a method to record any video
annotations within Flex application
UNIVERSITY COLLEGE DUBLIN  DUBLIN CITY UNIVERSITY  TYNDALL NATIONAL INSTITUTE
• The Flex application currently allows the
coach to draw on a video, but it doesn’t
save any video annotations.
• By recording video annotations, this
opens up the possibility for coaches to
annotate videos and players could
playback video with or without annotations
. – no research value in this though
UNIVERSITY COLLEGE DUBLIN  DUBLIN CITY UNIVERSITY  TYNDALL NATIONAL INSTITUTE