Google Research

Rich features for perceptual quality assessment of UGC videos


Video quality assessment for User Generated Content (UGC) is an important topic in both industry and academia. Many quality metrics focus one aspect of the quality, which more or less affect their performance on predicting overall UGC quality. In this paper, we create a large scale dataset to investigate characteristics of generic UGC video quality. Besides the subjective dataset, we also proposed a DNN-based framework to comprehensively analyze video quality in content, technical quality, and compression aspects. Our model is able to provides quality scores as well as human-friendly quality indicators, bridge the gap from low level video signals to human perceptual quality. Experiment results showed that our models achieved state-of-the art correlation with Mean Opinion Scores (MOS).

Research Areas

Learn more about how we do research

We maintain a portfolio of research projects, providing individuals and teams the freedom to emphasize specific types of work