Estimation of Emotional Scenes from Lifelog Videos with Moving Average of Facial Expression Intensity
Tomoya Nishimura, Hiroki Nomiya, Teruhisa Hochin · 2021
Lifelog video which saves the situation of daily life as video data has been proposed. In recent years, we are able to save much video data of daily life. But it has been difficult to select necessary data from a large amount of accumulated data. This paper proposes retrieval system with facial expression with the aim of retrieving important scenes from lifelog video data efficiently. In this paper, the emotional scene is tried to be detected with moving average of facial expression intensity. After the detection of the emotional scenes for lifelog videos by the proposed method is conducted, the proposed method is evaluated by comparing the proposed method with the method of preceding research through an evaluation experiment. As a result of the experiment, it is showed that the retrieval performance was improved.