Paper Link | University Press Release
We are pleased to announce that research using the ub-MOJI dataset has been published in the journal Electronics.
In this study, we developed a deep learning model based on finger angle features to distinguish subtle differences in complex finger movements in Japanese fingerspelling.
Furthermore, by introducing an algorithm that automatically detects the start and end points of sign language motions, we significantly improved recognition performance at the word level. This achievement lays the foundation for a high-performance fingerspelling recognition system with practical applications.
Looking ahead, we aim to enhance both generalizability and accuracy by collecting data from more diverse participants and incorporating context-aware segmentation methods.
This research is expected to contribute to the development of educational support tools and sign language interpretation systems, thereby supporting the deaf and hard-of-hearing community.
Publication Details
@Article{electronics14153052,
AUTHOR = {Kondo, Tamon and Murai, Ryota and He, Zixun and Shin, Duk and Kang, Yousun},
TITLE = {Recognition of Japanese Finger-Spelled Characters Based on Finger Angle Features and Their Continuous Motion Analysis},
JOURNAL = {Electronics},
VOLUME = {14},
YEAR = {2025},
NUMBER = {15},
ARTICLE-NUMBER = {3052},
URL = {https://www.mdpi.com/2079-9292/14/15/3052},
DOI = {10.3390/electronics14153052}
}