Duration: July 2016 – July 2019

The Celtic-PLUS project VIRTUOSE focuses on virtualizing video services to enable flexible usage of different computing platforms for robust and scalable video delivery and analysis.

In the German part of the project, VIRTUOSE-DE funded by the BMWI, we aim to develop low-complexity, real-time algorithms for analysis of large-scale visual data. The developed algorithms will be implemented on a cloud-based platform to provide flexibility and scalability for virtualized services. This platform will provide different solutions for “Video Analysis as a Service” (VAaaS) and will mainly target two use cases: Parking lot management and surveillance as well as smart on-street parking in the frame of the emerging smart-City concept, and video-based security in public transportation. To address these use cases, real-time action recognition and person/object tracking algorithms will be developed and deployed in a cloud environment. The focus will be on developing low-complexity algorithms that mainly operate on compressed video data to vastly reduce storage and processing requirements. Hierarchical approaches that combine deep learning-based computer vision techniques with compressed domain processing will be investigated.

Project consortium consists of 19 partners from Germany, Finland, Spain, Turkey and Romania.

VIRTUOSE Project partners:
Siemens AG, DResearch, Fraunhofer FOKUS, DCAITI/TU Berlin, VTT, Sofia Digital Ltd, Teleste, Tampere University of Technology, Tuxera, VideoVisit, BEIA Consult International, Nokia, I2CAT, Innovati Networks, Universidad Politecnica de Madrid, Koc University, NETAS, Vestel Electronics.

Project web site