Sumarização automática de vídeos em sessões de teste de usabilidade.

Abstract

On the context of product development, the usability evaluation stage is considered a complex and time consuming one. The time required to the analysis of each conducted test tends to be some times larger than the time required for the conduction of the test itself. Nevertheless, usability evaluation gathers the attention of an ever growing number of researchers. In every new research, the process of usability evaluation is either customized for a more specific case or more generic casts are created to the evaluation processes. Anyhow, a set of stages and recommendations remain getting propagated. It is the example of video recording. In such context, this research proposes a novel approach to automatic summarization of usability test session videos. Considering the analysis of such videos is generally conducted with little to none automation. The time required to finish the process is too long. Therefore, it is found the need to reduce the time associated to the analysis of usability test session videos. The approach described in this research seeks to offer summarized videos to the usability evaluator from the usability test session videos. The summarization process of this research is founded upon Active Appearance Models (AAM), and on a semantically adequate measure for the comparison of facial expressions. The validation of the developed approach consisted on the conduction of a usability test. The test was conducted in real work conditions. From the usability tests were extracted quantitative indicators. More indicators are extracted automatically by a system developed with the proposed approach. A comparative analysis of such indicators is the stepping stone for the validation of the proposed approach.