Video Super-Resolution
KEBULA is an innovative start-up founded as a spin-off from the University of Salerno, whose goal is to unlock the potential of data for future business. It helps Clients improve their skills in data engineering, analytics and AI. In addition, he is a founding member of MPAI, an organization that sets new standards for multimedia data coding using artificial intelligence. In the field of video streaming, the use of AI-based super-resolution models makes it possible to reduce the bit-rate required for the transmission of high-quality video, opening up new possibilities for many business applications.
Start date: 01/06/2021
Duration in months: 12
Problem Description
The use-case concerns the design of new video codecs that can reduce the bitrate required for the storage and transmission of high-quality video. Specifically of the development of a model for transmission and storage of low-resolution video, which can then be scaled to the desired resolution without significant loss of quality.
Goals
New services
Challenges
The challenge was to reduce the number of bit-rates required for storing and transmitting high-quality video.
Innovation results
The larger models (IconVSR and EDVR) recorded the best super-resolution quality, but could only run at 0.79 and 0.73 FPS on Marconi100, respectively, making them unsuitable for real-time applications.
Business impact
As KEBULA aims to cover a wide range of data analysis and transformation pipelines, experience on these kinds of models will be increasingly relevant to open up new business opportunities.Finally, since the PoC came about as part of the collaboration with MPAI to design new and improved video codecs enhanced by artificial intelligence algorithms, the knowledge gained from testing and evaluating all the models described will concretely help KEBULA determine possible paths to achieve this goal.