Grant: Engage Grants Program
École de technologie supérieure
$25,000.00
Adaptive Systems for Spatio-Temporal Recognition in Watchlist Screening Applications
FY 2016-17
Summary
This project focuses on adaptive systems that can accurately recognize individuals of interest in videosurveillance applications, and particularly in watchlist screening. Systems for still-to-video face recognition(FR) are employed in such applications, where they allow to capture facial trajectories across mobile anddistributed video cameras, and to compare them against stored facial models. The performance of these systemsis presently affected by the complex and changing operational environments, e.g., variations of facialappearance in pose and illumination. Moreover, since an individual is typically enrolled using one referencestills captured under controlled conditions, face models are not robust to intra-class variation. Faces capturedduring enrollment may differ considerably from those captured with operational surveillance cameras. Thisproject aims to provide Titans Unit Inc. with the expertise and engineering solutions needed to design robustFR systems that can adapt to different operational environments. In particular, to sustain a high level ofperformance, these spatio-temporal systems will adapt facial models to the operational domain based on newreference stills and videos. Robustness to different operational domains will be improved by generating diverseclassifier ensembles that exploit the abundance of non-target data captured from specific camera viewpoints.During operation, ensemble fusion functions will adapt over time for each facial trajectory, and facial modelswill be adapted based on new reference still and trajectories. In this project, the performance of advanceddomain adaptation and self-updating techniques that are suitable for watchlist screening applications will beevaluated and compared experimentally using anonymized video data provided by Titan Unit. Given the recentemergence of many powerful visual recognition techniques that can be applied in surveillance, the systemsdeveloped in this project will offer Titan Unit excellent business opportunities.
Program
Engage Grants Program
Recipient
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
499756-2016
Name nom
Granger, Eric
Department département
Génie de la production automatisée
Organization id
53
Institution établissement
École de technologie supérieure
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2016
Competition year année de concours
2016
Award amount
25000
Program id
EGP
Program name 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
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
2605
Research subject group en
Robotics
Research subject group fr
Robotique
Research subject en
Pattern analysis and machine intelligence
Research subject fr
Analyse de formes et intelligence artificielle
Partie
1
Nb partie
1
Application title
Adaptive Systems for Spatio-Temporal Recognition in Watchlist Screening Applications
Keyword
NSERC
Application summary
This project focuses on adaptive systems that can accurately recognize individuals of interest in videosurveillance applications, and particularly in watchlist screening. Systems for still-to-video face recognition(FR) are employed in such applications, where they allow to capture facial trajectories across mobile anddistributed video cameras, and to compare them against stored facial models. The performance of these systemsis presently affected by the complex and changing operational environments, e.g., variations of facialappearance in pose and illumination. Moreover, since an individual is typically enrolled using one referencestills captured under controlled conditions, face models are not robust to intra-class variation. Faces capturedduring enrollment may differ considerably from those captured with operational surveillance cameras. Thisproject aims to provide Titans Unit Inc. with the expertise and engineering solutions needed to design robustFR systems that can adapt to different operational environments. In particular, to sustain a high level ofperformance, these spatio-temporal systems will adapt facial models to the operational domain based on newreference stills and videos. Robustness to different operational domains will be improved by generating diverseclassifier ensembles that exploit the abundance of non-target data captured from specific camera viewpoints.During operation, ensemble fusion functions will adapt over time for each facial trajectory, and facial modelswill be adapted based on new reference still and trajectories. In this project, the performance of advanceddomain adaptation and self-updating techniques that are suitable for watchlist screening applications will beevaluated and compared experimentally using anonymized video data provided by Titan Unit. Given the recentemergence of many powerful visual recognition techniques that can be applied in surveillance, the systemsdeveloped in this project will offer Titan Unit excellent business opportunities.