🌍 Sign Language Dataset Hub

[] [] []

A curated, verified catalog of 73+ sign language datasets covering 26 sign languages — the most comprehensive open collection for sign language recognition (SLR) research, gesture recognition, deaf community technology, and assistive AI development.

---

🎯 Mission

To democratize access to sign language technology by providing:

• Verified dataset links — every URL checked, sample counts from original sources
• Working tools for loading, visualizing, and processing data
• Demo datasets for learning and prototyping
• Proper attribution to all data creators

Helping developers, researchers, and the deaf community build better assistive technology.

---

📊 Stats

Metric	Count
Verified Datasets	73
Sign Languages	26
Modalities	Video, Image, Sensor, Pose, RGB-D, Skeleton, Text
Source Verification	100% (all URLs checked)

Datasets by Language

Language	Code	Datasets	Notable
American Sign Language	ASL	11	MS-ASL, WLASL, How2Sign, OpenASL, ASLLVD
Arabic Sign Language	ArSL	2	ArSL2018, KArSL
Australian Sign Language	Auslan	1	Auslan Signbank
Bangla Sign Language	BdSL	4	BdSL47, Ban-Sign-Sent-9K
Brazilian Sign Language	Libras	2	Libras-UFPR, PHOENIX-Libras
British Sign Language	BSL	3	BOBSL, BSL Corpus, BSL SignBank
Chinese Sign Language	CSL	2	DEVISIGN, USTC-CSL
Dutch Sign Language	NGT	1	CNGT Corpus
French Sign Language	LSF	2	Dicta-Sign LSF, LSF-Dict
German Sign Language	DGS	3	RWTH-PHOENIX-2014, PHOENIX-2014T, DGS Corpus
Greek Sign Language	GSL	1	GSL-50
Indian Sign Language	ISL	3	INCLUDE, ISL-CSLTR, ISL-Alphabet
Irish Sign Language	ISL	1	ISL Corpus
Italian Sign Language	LIS	1	ATIS
Japanese Sign Language	JSL	1	J-ASL
Korean Sign Language	KSL	1	KETI
Malaysian Sign Language	BIM	1	MSL Dataset
Mexican Sign Language	LSM	1	LSM Sign Language
Russian Sign Language	RSL	2	RuSLAN, RSL-Signs
Swedish Sign Language	SSL	1	SSL Corpus
Thai Sign Language	TSL	1	TSL-51
Turkish Sign Language	TİD	1	AUTSL
Multilingual	—	5	SIGN-Hub, Dicta-Sign, SpreadTheSign, OpenSLR, SLP Toolkit
Linguistic DBs	—	6	ASL-LEX, BSL SignBank, Auslan Signbank, etc.

Datasets by Modality

Modality	Count
Video	35+
Image	10+
Video + RGB-D + Skeleton	3
Sensor (IMU/Flex)	1
Linguistic / Dictionary	6+
Multilingual Corpus	5+

---

✅ Verification Policy

All dataset source URLs in this repo have been verified. This means:

• ✅ Every URL resolves (HTTP 200 or auth-gated page)
• ✅ Sample counts are from the original publication, not estimated
• ✅ Every dataset has a citation or credit to its creator
• ✅ No placeholder or fabricated links exist in the active catalog
• ✅ Datasets we couldn't verify are excluded (not listed with fake info)

Found a broken link? Please open an issue.

---

📚 Browse Datasets

See DATASETS.md for the complete verified catalog with:

• 67 datasets across 22 sign languages
• Source URLs, sample counts, licenses, and citations for every entry
• Organized alphabetically by language

---

📖 Literature & Benchmarks

• REFERENCES.md — Key papers on sign language recognition, translation, datasets, and pose estimation
• docs/BENCHMARKS.md — Published accuracy numbers from real papers (WER, BLEU, accuracy)

---

🚀 Quick Start

Clone & Setup

git clone https://github.com/rudra496/SignLanguage-Dataset-Hub.git
cd SignLanguage-Dataset-Hub
pip install -r requirements.txt

Use the Demo Data (Bangla Sign Language Sensor Data)

from scripts.data_loader import BdSLSensorGloveDataset

# Load demo sensor data (4,824 samples)
dataset = BdSLSensorGloveDataset(split='train')
print(f"Loaded {len(dataset)} samples, 36 gesture classes")

sample = dataset[0]
print(f"Gesture: {sample['gesture_name']}")
print(f"Sensors shape: {sample['sensors'].shape}")

Visualize Sensor Data

python tools/visualize.py --data data/bdsl/BdSL-Sensor-Glove/

Browse Programmatically

import pandas as pd
df = pd.read_csv('datasets_catalog.csv')

# Filter by language
asl = df[df['language_code'] == 'ASL']
print(asl[['name', 'samples', 'source_url']])

# Filter by modality
video = df[df['modality'].str.contains('Video')]
print(f"Video datasets: {len(video)}")

Download External Datasets

# From Kaggle (requires Kaggle API key)
pip install kaggle
kaggle datasets download -d datamunge/sign-language-mnist
kaggle datasets download -d grassknoted/asl-alphabet
kaggle datasets download -d ahmedkhan123/arabic-sign-language

# From Hugging Face
pip install datasets
python -c "from datasets import load_dataset; ds = load_dataset('banglagov/Ban-Sign-Sent-9K-V1')"

# From Zenodo
wget https://zenodo.org/record/7067906/files/BdSL47.zip

---

🛠️ Included Tools

Tool	Description	Location
Data Loader	PyTorch dataset classes for sensor data	scripts/data_loader.py
Download Script	Multi-source dataset downloader	scripts/download_datasets.py
Visualizer	Sensor data visualization	tools/visualize.py
Data Generator	Demo data creation utilities	tools/generate_realistic_data.py

---

📁 Repository Structure

SignLanguage-Dataset-Hub/
├── DATASETS.md              # Complete verified dataset catalog (67 datasets)
├── datasets_catalog.csv     # Machine-readable catalog
├── STATISTICS.md            # Detailed statistics & breakdowns
├── README.md                # This file
├── CHANGELOG.md             # Version history
├── data/
│   └── bdsl/
│       └── BdSL-Sensor-Glove/  # Demo sensor dataset (4,824 samples)
├── docs/
│   ├── BENCHMARKS.md           # Published accuracy numbers & comparisons
│   ├── LICENSE_ATTRIBUTION.md  # Per-dataset license & citation info
│   ├── TUTORIALS.md            # 9 tutorials (beginner to advanced)
│   ├── QUICKSTART.md           # Quick start guide
│   └── CONTRIBUTING.md         # How to contribute
├── scripts/
│   ├── data_loader.py       # PyTorch data loaders
│   └── download_datasets.py # Multi-source downloader
├── tools/
│   ├── visualize.py         # Sensor data visualization
│   └── generate_realistic_data.py  # Data generation
├── .github/                 # Issue templates & PR template
├── CITATION.cff             # Citation metadata
├── LICENSE                  # CC BY 4.0
└── requirements.txt         # Python dependencies

---

📖 Tutorials

We include 9 tutorials from beginner to advanced:

#	Tutorial	Level
1	Introduction to Sign Language Recognition	Beginner
2	Loading and Exploring Datasets	Beginner
3	Visualizing Sign Language Data	Beginner
4	Building Your First Classifier	Intermediate
5	Hand Pose Estimation with MediaPipe	Intermediate
6	Data Augmentation Techniques	Intermediate
7	Real-time Recognition System	Advanced
8	Continuous Sign Language Recognition	Advanced
9	Multilingual Sign Recognition	Advanced

See docs/TUTORIALS.md and docs/QUICKSTART.md.

---

📚 Citation

If you use this repository, please cite:

@misc{signlanguage_dataset_hub,
  title     = {Sign Language Dataset Hub: A Verified Catalog of Sign Language Datasets},
  author    = {Sarker, Rudra and Contributors},
  year      = {2026},
  url       = {https://github.com/rudra496/SignLanguage-Dataset-Hub}
}

Please also cite the original dataset creators when using their data. See docs/LICENSE_ATTRIBUTION.md for per-dataset citation information.

---

🤝 Contributing

We welcome contributions! See docs/CONTRIBUTING.md.

---

🤔 Why This Repo?

Feature	This Repo	Typical SLR Papers/GitHub Lists	Kaggle Collections
Datasets	73+ curated	5–20 mentioned inline	10–30, unverified
Sign Languages	26	1–5	3–10
URL Verification	✅ All checked	❌ Often broken links	❌ Mixed
Sample Counts	From original sources	Inconsistent	User-reported
License Info	✅ Per dataset	Rarely included	Rarely included
Modality Tags	✅ All datasets	Partial	Tags vary
Tools & Scripts	✅ Included	❌	❌
Demo Datasets	✅ Included	❌	❌
Open Source	CC BY 4.0	Varies	Varies
Actively Maintained	✅	Usually one-time	Community

Rules:

1. Every dataset must have a verifiable source URL
2. Sample counts must come from the original source
3. Include license and citation information
4. No placeholder or fabricated data — ever

---

📖 Citation

If you use this dataset catalog in your research, please cite:

@misc{signlanguage_dataset_hub,
  title   = {Sign Language Dataset Hub: A Curated Catalog of 73+ Verified Datasets for 26 Sign Languages},
  author  = {Sarker, Rudra},
  year    = {2025},
  url     = {https://github.com/rudra496/SignLanguage-Dataset-Hub},
  note    = {Version 1.0}
}

---

📄 License

This repository is licensed under CC BY 4.0.

Individual datasets have their own licenses — see docs/LICENSE_ATTRIBUTION.md for details. Some datasets are research-use only and may require institutional agreements.

---

🙏 Acknowledgments

This hub would not be possible without the researchers and organizations who created and shared these datasets:

• Microsoft Research (MS-ASL)
• Oxford VGG (BOBSL)
• RWTH Aachen University (PHOENIX)
• Boston University (ASLLVD)
• IISc Bangalore (INCLUDE)
• CASIA / USTC (Chinese SL datasets)
• Ankara University (AUTSL)
• University of Hamburg (DGS Corpus)
• UCL DCAL (BSL Corpus, BSL SignBank)
• Stockholm University (SSL Corpus)
• Radboud University (CNGT)
• European Sign Language Centre (SpreadTheSign)
• Macquarie University (Auslan Signbank)
• EU Framework Programme (SIGN-Hub, Dicta-Sign)
• Hugging Face community datasets
• Kaggle community datasets
• And many more individual researchers and Deaf community members

---

📞 Contact

• Maintainer: Rudra Sarker
• Institution: Shahjalal University of Science and Technology

---

🌐 Connect

---

Built with ❤️ by Rudra Sarker
_{CC BY 4.0 License · Free & Open Source Forever}