Politecnico di Milano and EPFL have recently released a publicly available database to support reproducible research in the field of video quality assessment, specifically targeting H.264/AVC video streaming over error-prone networks.
Multimedia contents are often distributed over best-effort networks, i.e. there is no guarantee that the content will be delivered without errors to the final users. Therefore, service providers are interested in automatically monitoring the quality of experience at the receiver side, where the original content is typically unavailable. To this end, no-reference objective quality metrics aim at estimating the perceived quality, based on information that can be extracted from the received data. In order to validate the performance of any metrics, there is the need of comparing the results with subjective scores given by individuals, which are asked to watch and rate selected video contents. Collecting such data is time consuming and requires careful design of the subjective tests, in terms of test material, test methods and methodologies for processing subjective data.
Subjective tests have been carried out following state-of-the-art guidelines at the premises of two institutions, Politecnico di Milano and EPFL, in the context of the EU-funded VISNET-II project. An exhaustive database, including test material and subjective scores, have been made available to the research community. The database is available for download at the following links: