๐Ÿ  Back to Home

International Sign Language Collection

This repository documents International Sign Language Collection and multilingual sign language benchmark datasets for researchers. Our goal is to support multilingual, cross-cultural sign language understanding and generation by leveraging diverse large-scale datasets.

โœจ Overview

The datasets listed here are publicly documented resources commonly used in sign language understanding, recognition, translation, and generation research. Access conditions vary: some are directly downloadable, while others require agreements, institutional access, or author requests.

In addition, we organize a collection of open-access sign language benchmark datasets, covering major sign languages across the world.

๐Ÿ‡ฉ๐Ÿ‡ช German ๐Ÿ‡บ๐Ÿ‡ธ American ๐Ÿ‡ท๐Ÿ‡บ Russian ๐Ÿ‡ฆ๐Ÿ‡ท Argentinian ๐Ÿ‡ฌ๐Ÿ‡ง British ๐Ÿ‡น๐Ÿ‡ท Turkish ๐Ÿ‡ฆ๐Ÿ‡บ Australian ๐Ÿ‡ง๐Ÿ‡ท Brazilian ๐Ÿ‡ฎ๐Ÿ‡ณ Indian ๐Ÿ‡ซ๐Ÿ‡ท French ๐Ÿ‡ฐ๐Ÿ‡ท Korean ๐Ÿ‡ณ๐Ÿ‡ฑ Dutch / NGT ๐Ÿ‡ง๐Ÿ‡ช Flemish ๐Ÿ‡จ๐Ÿ‡ณ Chinese

๐Ÿ“Œ Whatโ€™s inside

  • Curated benchmark datasets with official/project/paper links
  • Coverage across multiple regions and sign languages
  • Dedicated section for the new dataset and its sources
Important Note: This Collection does not contain any New Zealand Sign Language (NZSL) data.

๐Ÿ†• International Sign Language Collection

๐Ÿ” Data Sources

The data sources used in this collection are publicly documented resources made available through official dataset pages, project websites, repositories, paper pages, or data access portals. Please follow the original licences and access conditions for each dataset.
Important Note: This Collection does not contain any New Zealand Sign Language (NZSL) data.

๐ŸŽฏ Purpose

  • Expand coverage to under-represented sign languages
  • Support future research on sign language understanding

๐Ÿ“Š International Sign Language Collection

The table below summarizes the sign language datasets curated in this repository, together with their languages, regions, approximate scale, and access links. *Please note that some datasets require contacting the original authors and obtaining permission before use.

Dataset Language Region #Samples #Class/Vocab #Signers URL
PHOENIX-2014 German Sign Language (DGS) ๐Ÿ‡ฉ๐Ÿ‡ช Germany 6,841 1,558 9
MS-ASL American Sign Language (ASL) ๐Ÿ‡บ๐Ÿ‡ธ USA 25,513 1,000 222
How2Sign American Sign Language (ASL) ๐Ÿ‡บ๐Ÿ‡ธ USA 80+ hrs 16K 11
Slovo Russian Sign Language (RSL) ๐Ÿ‡ท๐Ÿ‡บ Russia 20,400 1,000 194
LSA64 Argentinian Sign Language (LSA) ๐Ÿ‡ฆ๐Ÿ‡ท Argentina 3,200 64 10
BOBSL British Sign Language (BSL) ๐Ÿ‡ฌ๐Ÿ‡ง UK 1.2M sentences / 1400+ hrs 78K 37
AUTSL Turkish Sign Language (TฤฐD) ๐Ÿ‡น๐Ÿ‡ท Turkey 38,336 226 43
Auslan-Daily Australian Sign Language (Auslan) ๐Ÿ‡ฆ๐Ÿ‡บ Australia 25,106 13,945 67
LIBRAS-UFOP Brazilian Sign Language (LIBRAS) ๐Ÿ‡ง๐Ÿ‡ท Brazil 2,800 56 5
FDMSE-ISL Indian Sign Language (ISL) ๐Ÿ‡ฎ๐Ÿ‡ณ India 40,033 2,002 20
Mediapi-RGB French Sign Language (LSF) ๐Ÿ‡ซ๐Ÿ‡ท France 1,230 - >10
NIASL2021 Korean Sign Language (KSL) ๐Ÿ‡ฐ๐Ÿ‡ท Korea 201,026 pairs 7,989 signs -
NGT Sign Language of the Netherlands (NGT) ๐Ÿ‡ณ๐Ÿ‡ฑ Netherlands Corpus / lexical resources - -
VGT Flemish Sign Language (VGT) ๐Ÿ‡ง๐Ÿ‡ช Flanders 140 hrs / 5TB 16,928+ signs 120
CSL-Daily Chinese Sign Language (CSL) ๐Ÿ‡จ๐Ÿ‡ณ China 20,654 2K 10
SP-10 / Spreadthesign-Ten Multilingual sign languages ๐ŸŒ Multilingual Up to 100 language pairs - -
YouTube-SL-25 25+ sign languages ๐ŸŒ Multilingual 3,000+ hrs - -

Tip: On small screens, scroll horizontally to view all columns.

๐Ÿš€ Research Goals

This repository supports research on:

  • ๐Ÿค– Sign language translation
  • ๐ŸŽฅ Sign language generation
  • ๐ŸŒ Multilingual and cross-lingual modeling
  • ๐Ÿง  Inclusive AI for accessibility

Our work focuses on improving efficacy, generalization, and cultural coverage of sign language AI systems.

๐Ÿ“Ž Citation

If you find this repository useful in your research, please consider citing our work (coming soon).

coming soon
๐Ÿ  Back to Home