Grant: Discovery Grants Program - Individual
University of British Columbia
$14,000.00
Analysis and Interpretation of Visual Motion
FY 2014-15
Summary
The goal is to make machines see. The current focus is on the analysis and interpretation of motion. The objective is to allow a mobile robot to interact autonomously in increasingly complex environments. Object recognition often is taken to be the central problem in computer vision. Evidence from biological vision suggests that motion is primitive. The research questions are: What are the low-level motion features to detect? How can the tracking of low-level features support recognition (rather than follow it)? For each task, we begin with the question, ""What information about the 3D world is needed?"" This naturally leads to two other questions: 1) ""What limits our ability to acquire this information?"" and 2) ""What can we do with this information once we have it?"" One task is the analysis of hockey broadcast TV video. Limitations have been relaxed. We accommodate cameras that move (pan/tilt/zoom). We have improved our ability to transform the image coordinates of each video frame to (absolute) rink coordinates. New abilities are demonstrated. We determine puck possession automatically both directly, by tracking the puck when it is visible, and indirectly, via motion field analysis, when it is occluded. Team sports involve both cooperative and competitive behaviour. In each specific sport, there are well understood objectives which allow assumptions to be make about player intentions. The success or failure of specific behaviours also can be assessed. The research is an important step toward generalizable, motion-based ""situational intelligence."" On the internet, video is ""big data."" Exploiting knowledge about the world embedded in video leads to issues of what formal language to use to represent the needed information and how to compute this information from a video stream. Ultimately, we want to understand agent behaviour, both to provide an account of what has so far been observed and, more importantly long-term, to predict future behaviour.
Program
Discovery Grants Program - Individual
Recipient
Location
British Columbia, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
3390-2012
Name nom
Woodham, Robert
Department département
Computer Science
Organization id
2
Institution établissement
University of British Columbia
Province en
British Columbia
Province fr
Colombie-Britannique
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2014
Competition year année de concours
2012
Award amount
14000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
1507
Committee name en
Computer Science
Committee name fr
Sciences informatiques
Area of application code
1207
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Information, computer and communication technologies
Area of application fr
Technologie de l'information, des ordinateurs et des réseaux de communications
Research subject code
2603
Research subject group en
Robotics
Research subject group fr
Robotique
Research subject en
Computer vision
Research subject fr
Vision artificielle
Partie
3
Nb partie
5
Application title
Analysis and Interpretation of Visual Motion
Keyword
NSERC
Application summary
The goal is to make machines see. The current focus is on the analysis and interpretation of motion. The objective is to allow a mobile robot to interact autonomously in increasingly complex environments. Object recognition often is taken to be the central problem in computer vision. Evidence from biological vision suggests that motion is primitive. The research questions are: What are the low-level motion features to detect? How can the tracking of low-level features support recognition (rather than follow it)? For each task, we begin with the question, ""What information about the 3D world is needed?"" This naturally leads to two other questions: 1) ""What limits our ability to acquire this information?"" and 2) ""What can we do with this information once we have it?"" One task is the analysis of hockey broadcast TV video. Limitations have been relaxed. We accommodate cameras that move (pan/tilt/zoom). We have improved our ability to transform the image coordinates of each video frame to (absolute) rink coordinates. New abilities are demonstrated. We determine puck possession automatically both directly, by tracking the puck when it is visible, and indirectly, via motion field analysis, when it is occluded. Team sports involve both cooperative and competitive behaviour. In each specific sport, there are well understood objectives which allow assumptions to be make about player intentions. The success or failure of specific behaviours also can be assessed. The research is an important step toward generalizable, motion-based ""situational intelligence."" On the internet, video is ""big data."" Exploiting knowledge about the world embedded in video leads to issues of what formal language to use to represent the needed information and how to compute this information from a video stream. Ultimately, we want to understand agent behaviour, both to provide an account of what has so far been observed and, more importantly long-term, to predict future behaviour.