Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Discovery Grants Program - Individual

Université de Montréal

$20,000.00

Learning to represent images and their relations

EF 2013-14

Résumé
Finding a good data representation is a central first step in many intelligent information processing tasks, such as recognizing objects or faces in images. Recent advances in machine learning, known as deep learning and feature learning, have made it possible to automatically learn useful representations from raw data, and they have shown the potential to improve recognition performance in a variety of applications. Since feature learning methods extract representations from isolated, independent observations, they are complementary to traditional computer vision methods, like geometric inference methods, optical flow, or motion processing, which represent structure across multiple images not structure within a single image. This research will develop approaches to learning representations of the relationships between images. These may include images from multiple views of a scene or from neighboring frames in a video. Learning about relations will make it possible to introduce advances from feature learning into a broader set of computer vision tasks where relations are of interest. It will also provide a new, statistical perspective on motion processing, geometric inference and invariant object recognition. Applications of this research include advances in automated video understanding, image and video search on the web, and automatic multi-camera surveillance. New insights on the efficient encoding of image relations will also provide insights and testable hypotheses regarding theories of motion processing and geometric inference in the brain.
Programme
Discovery Grants Program - Individual
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
435717-2013
Name nom
Memisevic, Roland
Department département
Informatique et recherche opérationnelle
Organization id
63
Institution établissement
Université de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2013
Competition year année de concours
2013
Award amount
20000
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
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
5
Application title
Learning to represent images and their relations
Keyword
NSERC
Application summary
Finding a good data representation is a central first step in many intelligent information processing tasks, such as recognizing objects or faces in images. Recent advances in machine learning, known as deep learning and feature learning, have made it possible to automatically learn useful representations from raw data, and they have shown the potential to improve recognition performance in a variety of applications. Since feature learning methods extract representations from isolated, independent observations, they are complementary to traditional computer vision methods, like geometric inference methods, optical flow, or motion processing, which represent structure across multiple images not structure within a single image. This research will develop approaches to learning representations of the relationships between images. These may include images from multiple views of a scene or from neighboring frames in a video. Learning about relations will make it possible to introduce advances from feature learning into a broader set of computer vision tasks where relations are of interest. It will also provide a new, statistical perspective on motion processing, geometric inference and invariant object recognition. Applications of this research include advances in automated video understanding, image and video search on the web, and automatic multi-camera surveillance. New insights on the efficient encoding of image relations will also provide insights and testable hypotheses regarding theories of motion processing and geometric inference in the brain.