Embedded intention scripts representation and real-time interpretation metrics extraction methodology with gaze annotation on visual content

摘要

This paper proposes a system platform with a capability of analyzing and measuring the degree of similarities and differences of the gaze behaviors of the multiple users with the intention of the provider in real-time. In the proposed system platform, a video content incorporates the intention flow script, which describes the intended sequence and duration of the information embedded into the visual content. The video content also incorporates the intention weight script for assigning importance values to individual object information so that the significance of the gaze behavior can be assesses. In this paper, a set of evaluation metrics and their usages are defined, which can be used to evaluate the degree of gazing behavior closeness and deviation from the intention. The proposed method can be applied in many applications, such as in educational and commercial domains, where the information contents are usually designed with a specific intent. In the paper, we verified the proposed method by incorporation intention script into a sequence of educational image frames to analyze the gaze behavior of multiple users.