ub-MOJI logo ub-MOJI | AI Vision Lab

ub-MOJI: A Japanese Fingerspelling Video Dataset

ub-MOJI is a Japanese fingerspelling video dataset designed to advance research in sign language recognition

ub-MOJI Features

Japanese Fingerspelling Coverage

Supports syllables, 5-character sequences, and full words

Temporal Annotation

Precise start-end timing for every fingerspelled unit

Rich Participant Metadata

Includes detailed demographic and consent metadata for participant-aware modeling and analysis.

Academic-Only License

Available under terms restricting use to non-commercial academic research

Authors