Grant: Engage Grants Program
École Polytechnique de Montréal
$25,000.00
Video analytics for the identification of events of interest in user contexts
EF 2011-12
Résumé
An important role of surveillance and monitoring systems is to report events of interest to their users (emails,
SMS, etc.). A key requirement is that the system must reliably inform the user of all events of interest but at the
same time must not overload the user with useless notifications. Thus, events need to be filtered to keep only
the relevant ones. This is the specific problem that we are addressing in this project in order to improve the
usability of the surveillance system and extend its performance in event reporting. This is a very challenging
problem since what constitutes an event of interest varies from users to users. Currently, most commercial
systems simply give the users the possibility to interactively specify their preference by drawing zones and set
directionality, velocity, size parameters etc. This is a complex and time-consuming process and most users are
not ready to invest time in learning these software features.
Our investigations will therefore focus on researching methods and algorithms for the identification of events
of interest in user contexts. Our approach to solve this challenge is based on the following rationale. When a
detected event is identified as potentially of interest, the user is notified. This is then an opportunity to obtain
feedbacks. The user will be invited to rate the level of importance of the event and this information will be used
in the estimation of the 'interestness' of future events. This will be useful to tune the alerts for an individual
based on his responses. It will also allow broader, demographically or situationally (context) based adjustments
to be predictively made for new users or contexts. This is an anticipated area of further research.One of the key
for this learning to succeed, is to come up with an adequate representation of events. Since an event is
characterized by an actor (human, animal, or another object), motion, and an action, we will develop: 1) A
representation of an object's shape, appearance and velocity, 2) A trajectory representation, 3) A representation
of action in the spatio-temporal domain, and 4) A method to learn event of interest based on the user feedback.
Programme
Engage Grants Program
Bénéficiaire
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data
Tel que publié
Application id
412552-2011
Name nom
Bilodeau, GuillaumeAlexandre
Department département
Génie informatique et génie logiciel
Organization id
64
Institution établissement
École Polytechnique de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2011
Competition year année de concours
2011
Award amount
25000
Program id
EGP
Program nane en
Engage Grants Program
Program name fr
Programme de subventions d'engagement partenariat
Group en
Research partnerships programs
Group fr
Programmes de partenariat
Committee code
1552
Committee name en
Quebec Internal Decision Committee
Committee name fr
Comité de décision interne pour le Québec
Area of application code
802
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Information systems and technology
Area of application fr
Systèmes d'information et technologie de l'information
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
1
Nb partie
1
Application title
Video analytics for the identification of events of interest in user contexts
Keyword
NSERC
Application summary
An important role of surveillance and monitoring systems is to report events of interest to their users (emails,
SMS, etc.). A key requirement is that the system must reliably inform the user of all events of interest but at the
same time must not overload the user with useless notifications. Thus, events need to be filtered to keep only
the relevant ones. This is the specific problem that we are addressing in this project in order to improve the
usability of the surveillance system and extend its performance in event reporting. This is a very challenging
problem since what constitutes an event of interest varies from users to users. Currently, most commercial
systems simply give the users the possibility to interactively specify their preference by drawing zones and set
directionality, velocity, size parameters etc. This is a complex and time-consuming process and most users are
not ready to invest time in learning these software features.
Our investigations will therefore focus on researching methods and algorithms for the identification of events
of interest in user contexts. Our approach to solve this challenge is based on the following rationale. When a
detected event is identified as potentially of interest, the user is notified. This is then an opportunity to obtain
feedbacks. The user will be invited to rate the level of importance of the event and this information will be used
in the estimation of the 'interestness' of future events. This will be useful to tune the alerts for an individual
based on his responses. It will also allow broader, demographically or situationally (context) based adjustments
to be predictively made for new users or contexts. This is an anticipated area of further research.One of the key
for this learning to succeed, is to come up with an adequate representation of events. Since an event is
characterized by an actor (human, animal, or another object), motion, and an action, we will develop: 1) A
representation of an object's shape, appearance and velocity, 2) A trajectory representation, 3) A representation
of action in the spatio-temporal domain, and 4) A method to learn event of interest based on the user feedback.