Projektbeschreibung
Ein neues Zeitalter der Computergrafik einläuten
Die Technologie verändert fortwährend den Unterricht im Klassenzimmer und das Lernen im Allgemeinen. Von Online-Unterricht bis hin zu virtuellen Museumsführungen und einem verbesserten Gaming-Erlebnis – technologische Fortschritte erschließen neue Grenzen. In diesem Zusammenhang bringt das EU-finanzierte Projekt 3DIS-NN eine 3D-Bildsynthesetechnologie (3DIS) für die Wiedergabe von Objekten aus verschiedenen Perspektiven auf den Markt, um zahlreiche Anwendungen im Bereich der Computergrafik und Bilderkennung zu ermöglichen. Durch die Entflechtung von Objektattributen und deren Verflechtung über einen Renderer für die Synthese kann 3DIS eine Technologie bereitstellen, die hilfreiche Eigenschaften aus unserer visuellen Welt lernt. Diese können zum Verständnis von Videos genutzt werden, was mit zu den größten Zielen künstlicher Intelligenz zählt.
Ziel
3D (3-dimensional) Image Synthesis (3DIS) is a technology to render objects from different views which enables numerous applications in computer graphics and computer vision. As the digital world is becoming more crucial especially in the times of pandemic, 3DIS can provide tools for online classes, virtual social tours, improved gaming experience and simulators for robotics by providing realistic virtual 3D environments. Furthermore, 3DIS by disentangling the attributes of objects and entangling them via a renderer for synthesize, can provide a technology to learn useful features from our visual world that can be used for video understanding, one of the biggest goals of artificial intelligence. Here, I propose 3DIS-NN, a set of methods to improve the quality of 3DIS with deep neural networks (DNNs), and bring it close to the production quality, which will contribute to the European Union’s Future and Emerging Technology ambitions of Horizon Europe. Learning 3DIS from 2D images with deep learning is a challenging topic due to its inherent ambiguity. 3DIS-NN will enable high-quality 3DIS results by i) creating a dataset with weak labels to feed the data-hungry DNNs for better accuracy, ii) improving robustness of 3D geometry and texture prediction from images, iii) handling the impurities in segmentation of objects with a novel design of architecture, and iv) providing a tool to further close the domain gap in renderers and real images. This interdisciplinary proposal which is at the intersection of deep learning and computer graphics will be carried out at under the supervision of Prof. Ugur Gudukbay who is an expert in computer graphics. In terms of career developments, this proposal will consolidate and accelerate my career on the international landscape scene as a pioneer lead authority in the new cross-disciplinary area of “deep learning & computer graphics”.
Wissenschaftliches Gebiet
- natural sciencescomputer and information sciencesartificial intelligencecomputer vision
- natural sciencescomputer and information sciencesartificial intelligencemachine learningdeep learning
- natural sciencesmathematicspure mathematicsgeometry
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringrobotics
- natural sciencescomputer and information sciencesartificial intelligencecomputational intelligence
Schlüsselbegriffe
Programm/Programme
Thema/Themen
Aufforderung zur Vorschlagseinreichung
Andere Projekte für diesen Aufruf anzeigenFinanzierungsplan
MSCA-IF - Marie Skłodowska-Curie Individual Fellowships (IF)Koordinator
06800 Bilkent Ankara
Türkei