Self-made-60 Korean Sign Language Evaluation Dataset

Dataset Summary

This dataset contains 60 self-recorded Korean Sign Language isolated-word MP4 clips created by the project team for additional evaluation.

Contents

• videos/: 60 MP4 files
• File names are the target Korean word labels, for example: 간호사.mp4
• Local source folder before upload: /home/moon/Downloads/self-made-60
• Local folder size before upload: 370.4 MB

Intended Use

This dataset is intended for reproducible evaluation of the submitted Korean Sign Language isolated-word recognition pipeline. It is a small project-made evaluation set, not a general benchmark.

Label Format

The target label is the MP4 filename stem. Example:

videos/깨끗하다.mp4 -> 깨끗하다

Source And Licensing Notes

The videos were recorded by the project team. They are provided for course/project evaluation of the submitted model and inference pipeline.

AI Tools Used

AI tool used: Codex. Codex was used to package/upload the dataset and draft this dataset card. Codex was not used to create the videos or labels.