data analyst

Project description Objective quality assessment has the ultimate goal of providing a score which is highly correlated with the quality perceived by human observers. The topic of objective quality assessment is a well-studied and mature subject in the literature. As such, there are different proposals are available from the literature. Despite this, some of them fail to correlate entirely with subjective quality scores. And this may become particularly true when content containing text and graphics, such as gaming and desktop material, is considered. This claim gains more evidence if one considers that video quality metrics were primarily devised and/or parametrised to work on camera captured content (aka natural content). This project aims to assess and/or improve the prediction accuracy of a set of state-of-the-art quality metrics over video content which comprises natural, desktop and gaming content. The project will highlight pros and cons of each metric tested and point out different directions along which research can be devoted to improve these metrics. Useful resources: 1) Datasets: https://live.ece.utexas.edu/research/Quality/subjective.htm (Subjective database Release 2 ) https://live.ece.utexas.edu/research/ChallengeDB/index.html https://github.com/niu-haoran/FLIVE_Database/blob/master/database_prep.ipynb and many more: https://stefan.winkler.site/resources.html 2) Models (with code) https://live.ece.utexas.edu/research/Quality/index_algorithms.htm https://github.com/idealo/image-quality-assessment and many more: https://www.its.bldrdoc.gov/vqeg/downloads.aspx