Parmar Paritosh ; Morris Brendan Tran

Abstract

Can learning to measure the quality of an action help in measuring thequality of other actions? If so, can consolidated samples from multiple actionshelp improve the performance of current approaches? In this paper, we carry outexperiments to see if knowledge transfer is possible in the action qualityassessment (AQA) setting. Experiments are carried out on our newly released AQAdataset (http://rtis.oit.unlv.edu/datasets.html) consisting of 1106 actionsamples from seven actions with quality scores as measured by expert humanjudges. Our experimental results show that there is utility in learning asingle model across multiple actions.