岡田 龍太郎

Through technology, it is essential to seamlessly bridge the divide between diverse speaking communities (including the signer (the sign language speaker) community). In order to realize communication that successfully conveys emotions, it is necessary to recognize not only verbal information but also non-verbal information. In the case of signers, there are two main types of behavior: verbal behavior and emotional behavior. This paper presents a sign language recognition method by similarity measure with emotional expression specific to signers. We focus on recognizing the sign language conveying verbal information itself and on recognizing emotional expression. Our method recognizes sign language by time-series similarity measure on a small amount of model data, and at the same time, recognizes emotion expression specific to signers. Our method extracts time-series features of the body, arms, and hands from sign language videos and recognizes them by measuring the similarity of the time-series features. In addition, it recognizes the emotional expressions specific to signers from the time-series features of their faces.

In this paper, we present an emotion estimation method using heart rate variability parameters of vital data. Recently, as sensors have become more precise and smaller, it has been possible to obtain users' vital data in real-time quickly. In our method, ECG (electrocardiogram) data are measured beforehand while listening to a story with voice narration that evokes emotions and based on the trends obtained through the measurement, the emotions that have a high correlation with the newly acquired ECG data are estimated to be the emotions expressed in the ECG data. With the implementation of our method, it is possible to estimate the user's emotions based on ECG data. In this paper, we also represent the application of our method to chat icons that see users' emotions in real-time. By realizing this application, users will see the changes in their emotions and control their mental health.

本稿では，メディアコンテンツを対象とした統計的一般化逆作用素の構成方式を示す．これは，与えられた印象を表す単語とその重みで表される印象メタデータからメディアコンテンツを生成する機構を実現するものである．本方式は，従来，我々が提案してきた手法であるメディアコンテンツから印象を表す言葉をメタデータとして抽出する，メディアコンテンツを対象としたメタデータ自動抽出方式手法の逆演算として構成される．しかしながら，本逆演算においては一般的に，不良設定問題が発生する．この不良設定問題を解決するために，制約条件としてそのメディアの種類に応じた統計情報やそのメディアを対象とした研究成果・理論を用いる．これにより，メディアコンテンツと言葉の間の相互変換を可能とする．さらに，本方式を楽曲メディアコンテンツに適用し，印象語で表現された印象メタデータから楽曲メディアコンテンツを生成するシステムを実装する．これを用いて実験を行い，提案方式が実現されていることを示す．In this paper, we present a construction method of Stochastic generalized inverse operator for media contents. This method realizes automatic media contents creation depending on impression words as an inverse operation of our automatic metadata extraction method which we have proposed. This is an inverse operator which extracts words from media contents. However, this creation mechanism contains ill-posed problems. In order to solve them, we introduce stochastic data about the media and it's studies. Moreover, we construct a mutual conversion between impression words and media contents. Furthermore, we apply our method to music data creation system. We performed verification experiments, and showed the effectiveness of our method.