Sencogi Spatio-Temporal Saliency: A New Metric for Predicting Subjective Video Quality on Mobile Devices Maria Laura Mele 1,2,3( ✉ ) , Damon Millar 1 , and Christiaan Erik Rijnders 1 1 COGISEN Engineering Company, Rome, Italy {marialaura,damon,chris}@cogisen.com 2 Department of Philosophy, Social and Human Sciences and Education, University of Perugia, Perugia, Italy 3 ECONA, Interuniversity Centre for Research on Cognitive Processing in Natural and Artificial Systems, Sapienza University of Rome, Rome, Italy Abstract. Objective Video Quality Assessment (VQA) is often used to predict users visual perception of video quality. In the literature, the performance eval‐ uation of objective measures is based on benchmark subjective scores of perceived quality. This paper shows the evaluation of an algorithmic measure on videos presented on mobile devices. The VQA measure is called Sencogi Spatio- Temporal Saliency Metric (Sencogi-STSM), and it uses a spatio-temporal sali‐ ency to model subjective perception of video quality. Since STSM was previously validated with a subjective test conducted on laptop computers, the goal of this work was to verify whether the measure is able to significantly predict users’ perception of video quality also on mobile devices. Results show that, compared to the standard VQA metrics, only Sencogi-STSM is able to significantly predict subjective DMOS. This paper describes Sencogi-STSM’s biologically plausible model, its performance evaluation and the comparison with the most commonly used objective VQA metrics. Keywords: Video quality perception · Computer vision Spatio-temporal saliency · Objective video quality assessment 1 Quality Assessment Today 1.1 Introduction Subjective testing with human participants is still the most reliable method to assess the perceived quality of an image or a video, even though it requires high cost and time effort. In order to measure participants’ opinion scores of quality, standard recommen‐ dations are currently used, but the whole evaluation procedure often results in time- consumption and requires high costs. To avoid the cost and delay of subjective quality evaluation, objective quality assessment methodologies that do not involve participants are proposed in the literature to be used instead of subjective tests. © Springer International Publishing AG, part of Springer Nature 2018 M. Kurosu (Ed.): HCI 2018, LNCS 10902, pp. 552–564, 2018. https://doi.org/10.1007/978-3-319-91244-8_43