Keishi Tainaka

Keishi Tainaka

Auto-generated profile, based on publicly available information

Keishi Tainaka

KDDI Research, Inc., Fujimino, Saitama, Japan

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
10 December 2022 in Computers in Industry
Verified
Journal Article
Open access
KTKeishi Tainaka
Keishi Tainaka
YFYuichiro Fujimoto
Yuichiro Fujimoto
TSTaishi Sawabe
Taishi Sawabe
MKMasayuki Kanbara
Masayuki Kanbara
HKHirokazu Kato
Hirokazu Kato

9 April 2025 in IEEE Access
Verified
Journal Article
Open access
KTKeishi Tainaka
Keishi Tainaka
THToshiharu Horiuchi
Toshiharu Horiuchi
HPHanif Fermanda Putra
Hanif Fermanda Putra
TKTatsuya Kobayashi
Tatsuya Kobayashi
Embodied Conversational Agents (ECAs) can simulate human behaviors and express natural emotions. To further enhance these emotional expressions, we focused on applying attractive anime effects to the ECAs. Anime effects are visual techniques mainly used in anime productions, encompassing a wide range of unique types and combinations. Among these, we focused on anime effects that indirectly support and enhance the emotions expressed by anime characters. If properly applied to ECAs, it is expected to dramatically enhance their ability to express emotion. However, the absence of established methods to apply diverse anime effects to ECAs has constrained their emotional expressiveness. Therefore, we proposed a system that automatically applies these diverse anime effects to the ECAs. First, we surveyed and classified anime effects applicable to ECAs and created an anime effect dataset. Utilizing the dataset, we developed a system that automatically applies diverse anime effects to an ECA according to the ECA’s profile (gender and personality) and emotions. The proposed system was evaluated with 21 participants using two static use cases based on watching videos and with eight participants using two interactive use cases using digital signage. The results showed that the proposed system enhanced the intensity of emotional expression of the ECAs, suggesting an improvement in their interactions with users. Additionally, the applied anime effects allowed users to perceive the differences in personality between the two ECAs.

28 November 2023
Verified
Conference Paper
Open access
SUSoshiro Ueda
Soshiro Ueda
KTKeishi Tainaka
Keishi Tainaka
Yiming ShenYiming Shen
Yiming Shen
SUShuntaro Ueda
Shuntaro Ueda
KKKonstantin Kulik
Konstantin Kulik
HKHirokazu Kato
Hirokazu Kato
Augmented reality (AR) task support systems have not yet been widely spread in actual companies despite the fact that many researchers have worked tirelessly to validate their effectiveness for various industrial tasks. One of the reasons for this is the excessive development cost of a practical AR system. Development still strongly depends on outstanding developers’ skills and trial-and-error. To solve this problem, we propose to separate an AR task support system into a reusable software part (including basic functions and its user interfaces) and task support contents that differ depending on each task. We have developed a general software platform and an original file format for describing task support contents based on analyzing various tasks. In addition, we have also proposed a web tool that enables easy design of AR task support content even for those without AR development experience. The output of the tool, an XML file written in the format, is loaded into the software platform to enable operators to use the AR task support system. We demonstrate a series of steps from the creation of task support content to the use of the AR task support system.