Machine-learning based Blind Visual Quality Assessment with Content-aware Data Partitioning

Abstract

Over the years different machine-learning based image quality assessment models have been proposed. In this paper, we analyze data partitioning. Since statistical data partitioning may affect the results due to the number of iterations, we analyze the effect of content-aware partitioning. The results are analyzed for different partitioning methods and models using publicly available dataset and difference mean opinion scores.