SignLanguage-Dataset-Hub

Sign Language Research - Key References

A curated list of the most important papers in sign language recognition, translation, and production. Links are verified where possible; arxiv IDs marked with ✓ have been confirmed.

---

Survey Papers

1. Deep Learning for Sign Language Recognition: Current Techniques, Benchmarks, and Open Issues
   • Al-Qurishi et al., 2021, IEEE Access
   • IEEE
   • Review of machine/deep learning methods for SLR with benchmark analysis. Cited 250+.
2. A Comprehensive Study on Deep Learning-based Methods for Sign Language Recognition
   • Adaloglou et al., 2021, IEEE Trans. Image Processing
   • arXiv:2007.12530 ✓
   • Comparative assessment of computer vision-based SLR methods on multiple datasets.
3. Sign Language Recognition: A Comprehensive Review of Traditional and Deep Learning Approaches, Datasets, and Challenges
   • Tao et al., 2024, IEEE Access
   • IEEE
   • Comprehensive review covering traditional and DL methods.
4. Deep Learning Approaches for Continuous Sign Language Recognition: A Comprehensive Review
   • Khan et al., 2025, IEEE Access
   • IEEE
   • Focused review on continuous SLR methods and datasets.
5. Survey on Sign Language Recognition in Context of Vision-Based and Deep Learning
   • Subburaj & Murugavalli, 2022, Measurement: Sensors
   • Elsevier
   • Classification strategies from early HMM to hybrid and DL techniques.
6. From Rule-Based Models to Deep Learning Transformers for NLP and Sign Language Translation: Survey, Taxonomy and Performance Evaluation
   • 2024, arXiv
   • Broad survey covering SLT from rule-based to transformer approaches.
7. A Comparative Analysis of Techniques and Algorithms for Recognising Sign Language
   • Kumar et al., 2023, arXiv
   • Comparison of various SLR techniques and algorithms.
8. Sign Language Production: A Review
   • Rastgoo et al., 2021, Int. J. Comput. Vis.
   • arXiv
   • Comprehensive review of sign language production methods.
9. All You Need In Sign Language Production
   • Rastgoo et al., 2022, IEEE TPAMI
   • arXiv
   • Extended survey covering full pipeline of sign language production.
10. A Critical Study of Recent Deep Learning-Based Continuous Sign Language Recognition
    • Taher & Zeebaree, 2025, Review of Socionetwork Strategies
    • Focused study on recent advances in continuous SLR.

---

Datasets & Benchmarks

PHOENIX-2014 / PHOENIX-2014T

• ** RWTH-PHOENIX-Weather 2014: A Large Video Dataset for German Sign Language Recognition and Translation**
  • Forster et al., 2014, ICMI
  • DOI
  • The seminal continuous SLR dataset (DGS, weather broadcasts). 2014T adds German translations for SLT.

MS-ASL

• MS-ASL: A Large-Scale Data Set and Benchmark for Understanding American Sign Language
  • Joze et al., 2019, arXiv
  • arXiv:1812.01053 ✓
  • ~25k videos, 1000 classes of ASL signs from YouTube.

WLASL

• Word-Level American Sign Language Dataset (WLASL)
  • Li et al., 2020, Int. J. Comput. Vis.
  • Over 2000 glosses, 21k videos from multiple sources.

How2Sign

• How2Sign: A Large-Scale Multimodal Dataset for Continuous American Sign Language
  • Duarte et al., 2021, CVPR
  • CVF
  • ~80 hours, multimodal (RGB, pose, depth), continuous ASL with gloss annotations.

AUTSL

• AUTSL: A Large Scale Multi-Modal Turkish Sign Language Dataset and Baseline Methods
  • Sincan & Keles, 2020, IEEE Access
  • IEEE
  • 36k+ videos, 226 signs, 43 signers, multi-modal (RGB + depth + skeleton).

BOBSL

• BBC-Oxford British Sign Language Dataset
  • Albanie et al., 2021, arXiv
  • arXiv:2111.03635 ✓
  • Large-scale BSL dataset from BBC programming with subtitles.

SDW-ASL

• SDW-ASL: A Dynamic System to Generate Large Scale Dataset for Continuous American Sign Language
  • Jiang, 2022, arXiv
  • arXiv:2210.06791 ✓
  • Large-scale continuous ASL dataset generation system.

ASLLVD

• A Large Lexicon Dataset for Continuous American Sign Language
  • Boston University
  • Over 3000 signs from native ASL signers.

CSL-Daily

• CSL-Daily: A Large-Scale Chinese Sign Language Dataset and Benchmark
  • Zhou et al., 2021, CVPR
  • 20k+ sentences, 200 glosses, 30 signers of Chinese Sign Language.

INCLUDE

• INCLUDE: A Large-Scale Dataset for Sign Language Recognition
  • Project page
  • Large-scale sign language dataset with multiple modalities.

---

Recognition Methods

Early Approaches

• Recognition of Isolated Sign Language Gestures Using Hidden Markov Models
  • Vogler & Metaxas, 1998, FG
  • Pioneering HMM-based approach for SLR using 3D motion features.
• American Sign Language Recognition Using Stochastic Trajectory Matching
  • Lin & Costello, 2001
  • Template matching approach for ASL recognition.

Deep Learning - CNN & RNN

• Sign Language Recognition with 3D Convolutional Neural Networks
  • Pigou et al., 2015, ICCV Workshops
  • Early application of 3D CNNs to sign language recognition.
• Two-Stream Convolutional Networks for Sign Language Recognition
  • Bilen et al., 2016, BMVC
  • RGB + optical flow two-stream approach for SLR.
• Deep Sign Language Recognition: CNN + BiLSTM
  • Koller et al., 2019, PAMI
  • Two-stage CNN + BLSTM with CTC for continuous SLR on PHOENIX.

Deep Learning - Transformers & Attention

• Better Sign Language Translation with STMC-Transformer
  • Yin & Read, 2020, COLING
  • arXiv:2004.00588 ✓
  • Spatiotemporal multi-head attention for gloss-to-text translation.
• SlowFast Network for Continuous Sign Language Recognition
  • Ahn et al., 2024, ICASSP
  • arXiv:2309.12304
  • SlowFast backbone for feature extraction in continuous SLR. Cited 60+.

Graph-Based Methods

• Spatiotemporal Graph Convolutional Networks for Sign Language Recognition
  • Various authors, 2020+
  • Skeletal graph-based approaches leveraging joint connectivity.
• Sigma: Semantically Informative Pre-training for Skeleton-based Sign Language Understanding
  • Pu et al., 2025, arXiv
  • Self-supervised pre-training for skeleton-based SLU tasks.

SOTA Methods

• Sign Language Transformers: Joint End-to-End Sign Language Recognition and Translation
  • Camgoz et al., 2020, CVPR
  • Joint SLR and SLT with transformer architecture. Groundbreaking end-to-end approach.
• VAC (Visual Alignment Constraint) for Continuous Sign Language Recognition
  • Various, 2021+
  • Alignment-based methods improving frame-level features for CTC.

---

Translation

Gloss-Based Approaches

• Neural Sign Language Translation
  • Camgoz et al., 2018, CVPR
  • CVF
  • First neural approach to SLT using CTC recognition + NMT translation.
• A Simple Multi-Modality Transfer Learning Baseline for Sign Language Translation
  • Chen et al., 2022, CVPR
  • CVF
  • Strong baseline using pre-trained features + transformer. Cited 270+.

End-to-End Approaches

• Sign Language Transformers: Joint End-to-End Sign Language Recognition and Translation
  • Camgoz et al., 2020, CVPR
  • End-to-end SLR+SLT without intermediate gloss supervision.
• Sign Language Translation with Pseudo-Glosses
  • Various, 2022+
  • Training without ground truth glosses using pseudo-gloss generation.
• Lost in Translation, Found in Embeddings: Sign Language Translation and Alignment
  • Jang et al., 2025, arXiv
  • Embedding-based alignment for robust SLT with zero-shot generalization.

Scaling

• Scaling Sign Language Translation
  • Zhang et al., 2024, NeurIPS
  • NeurIPS
  • Large-scale approaches to improving SLT performance.

---

Pose Estimation & Feature Extraction

• MediaPipe Hands: On-device Real-time Hand Tracking
  • Zhang et al., 2020, arXiv
  • Widely used for real-time hand landmark extraction in SLR systems.
• OpenPose: Realtime Multi-Person 2D Pose Estimation
  • Cao et al., 2019, IEEE TPAMI
  • Standard tool for body/hand pose extraction from sign language videos.
• HRNet: Deep High-Resolution Representation Learning for Visual Recognition
  • Sun et al., 2019, CVPR
  • High-resolution pose estimation used for feature extraction in SLR.
• SignVTCL: Multi-Modal Continuous SLR Enhanced by Visual-Textual Contrastive Learning
  • Chen et al., 2024, arXiv
  • Contrastive learning between visual features and text for improved SLR.
• Multi-Stream Keypoint Attention Network for SLR and Translation
  • Guan et al., 2024, arXiv
  • Keypoint-based multi-stream attention for SLR/SLT tasks.

---

Real-time & Deployment

• Real-Time Sign Language Detection Using Deep Learning
  • Selvi et al., 2025, IEEE
  • Real-time SLR system deployment.
• Real-Time American Sign Language Recognition Using 3D CNN and LSTM
  • Various, 2025, arXiv
  • Webcam-based real-time ASL recognition system.
• Tslformer: A Lightweight Transformer Model for Turkish Sign Language Recognition Using Skeletal Landmarks
  • Ertürk et al., 2025, arXiv
  • arXiv:2505.07890
  • Lightweight transformer on skeleton data for efficient SLR.
• Mobile/Edge Deployment for Sign Language Recognition
  • Various works exploring model compression, quantization, and mobile deployment for SLR.

---

Sign Language Production

• MaDiS: Taming Masked Diffusion Language Models for Sign Language Generation
  • Zuo et al., 2026, arXiv
  • Diffusion-based sign language production achieving state-of-the-art.
• M3T: Discrete Multi-Modal Motion Tokens for Sign Language Production
  • Symeonidis-Herzig et al., 2026, arXiv
  • Multi-modal motion tokens for SOTA sign language production.
• MS2SL: Multimodal Spoken Data-Driven Continuous Sign Language Production
  • Ma et al., 2024, Findings of ACL
  • ACL Anthology
  • Spoken language to continuous sign language production.

---

Resources

• sign-language-processing/sign-language-processing — PyTorch-based toolkit for sign language processing
• dpFBSLDP)) — Deep feature based SLR papers and code
• Awesome Sign Language Resources)) — Curated list of SL datasets, papers, and tools

---

Last updated: March 2026. This list is maintained alongside the dataset catalog. Contributions welcome via pull request.

This site is open source. Improve this page.