UkrSL: Towards a Ukrainian Continuous Sign Language Dataset

2026-06-28 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Expert, quick

Summary

UkrSL is a newly released, annotated dataset for Ukrainian Sign Language (USL), one of Europe's most underresourced sign languages. This initial snapshot comprises 1,456 annotated video clips, totaling approximately two hours of signing, with 1,463 segments including cropped videos. The data was sourced from six broadcast videos provided by Suspilne, Ukraine's public broadcaster. Each clip features a spoken Ukrainian transcription precisely aligned to its corresponding signing segment. The dataset's creators describe the data collection pipeline and annotation methodology, providing detailed statistics and outlining limitations. UkrSL is actively being expanded, and its release aims to support the research community and encourage collaborative efforts in sign language processing.

Key takeaway

For research scientists and NLP engineers developing models for sign language understanding, you should consider integrating the UkrSL dataset into your work. This resource offers crucial annotated data for Ukrainian Sign Language, a significantly underresourced domain. Utilizing UkrSL can advance model development for sign language translation and recognition, and you are invited to collaborate on its ongoing expansion to further enrich this vital linguistic resource.

Key insights

UkrSL provides a critical annotated dataset for the underresourced Ukrainian Sign Language, fostering research and collaboration.

Method

The dataset was created using a defined data collection pipeline and an annotation methodology, aligning spoken Ukrainian transcriptions with sign language segments from broadcast videos.

In practice

• Utilize UkrSL for sign language model training
• Contribute to dataset expansion efforts

Topics

• Ukrainian Sign Language
• Sign Language Processing
• Low-Resource Languages
• Dataset Annotation
• Natural Language Processing
• Video Datasets

Best for: AI Scientist, NLP Engineer, Research Scientist

Related on AIssential

• Bootstrapping Sign Language Annotations with Sign Language Models — A pseudo-annotation pipeline uses sparse recognizer predictions and LLMs to generate sign language annotations.
• Ukrainian Visual Word Sense Disambiguation Benchmark — Multimodal LLMs significantly underperform in Ukrainian Visual-WSD compared to English, revealing a critical language resource gap.
• Sign-Language Datasets at Scale: A Comprehensive Survey on Resources, Benchmarks, and Annotation Standards — Fragmented sign language datasets and inconsistent annotations hinder robust, scalable AI development.
• Libras-UFPel Corpus: A Parallel Dataset of Brazilian Sign Language and Portuguese for Multimodal Research and Processing — The Libras-UFPel Corpus provides a parallel multimodal dataset for Brazilian Sign Language and Portuguese.
• Lexicon-Grammar Web — A web-based interface provides interactive access to a large-scale European Portuguese grammatical database.
• mAceReason-Math: A Dataset of High-Quality Multilingual Math Problems Ready For RLVR — mAceReason-Math provides a high-quality, multilingual dataset for advanced RLVR in math and logic.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.