Giters

volley84 / sign-lanuage-datasets

These datasets are used for machine-learning research

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Sign-lanuage-datasets

These datasets are used for machine-learning research

Sign language datasets

  • Co: Country
  • Class: Classes
  • Subj: Subjects
  • LL: Language level(W-Word,S-Sentence,H-Handshape)
  • Type: Type(V-Video, VR-Video(RGB), VD-Video(depth))
  • An: Annotations
  • Av: Availability(CA-Contact Author, PA-Publicly Available, Un-Unknown, Non-Non available)
  • T: There is in our hard drive?(Y-Yes, N-No)
id Dataset name Co Class Subj Samples Data LL Type An Av T
1 DGS Kinect 40 Ger 40 15 3000 W V,[9] PA Y
2 RWTH-PHOENIX-Weather Ger 1200 9 45760 52gb S V [18] PA Y
3 SIGNUM Ger 450 25 33210 920gb S V PA,[5] N
4 GSL 20 Gre 20 6 ~840 W CA Y
5 Boston ASL LVD USA 3300+ 6 9800 W V,[9] [19,20] PA N
6 PSL Kinect 30 Pol 30 1 30×10=300 ~1.2gb W V,[10] PA Y
7 PSL ToF 84 Pol 84 1 84×20=1680 ~33gb W V,[11] PA N
8 PSL 101 Pol ? ? ? ? ? ? CA N
9 LSA64 Arg 64 10 3200 20gb W VR [21] PA Y
10 BosphorusSign Tur Non N
11 MSR Gesture 3D USA 12 10 336 28mb W VD PA N
12 DEVISIGN-G Chi 36[1] 8 432 ? W VR CA N
13 DEVISIGN-D Chi 500 8 6000 ? W VR CA N
14 DEVISIGN-L Chi 2000 8 24000 ? W VR CA N
15 IIITA -ROBITA Ind 23 ? 284mb W VR,[15] CA N
16 Purdue ASL USA ? 14[3] ? ? W/S V,[14] [6] N
17 CUNY ASL USA ? 8 ~33000[4] ? S VR,[16] [7] U N
18 SignsWorld Atlas Ara [2] 10 ? ? W,S,H V,[17,14] ? U N

[1] - letters/numbers; [2] - multiple types; [3] - only 5 available; [4] - glosses; [5] - 1TB, contact author to obtain hard drive; [6] - Request DVDs/HD; [7] - Signstream; [8] - ?; [9] - multiple angles; [10]- depth from Kinect camera; [11]- ToF camera; [12]- ?; [13]- ?; [14]- RGB; [15]- 320x240; [16]- mocap data; [17]- Images; [18]- Face, hand, end/start(unfinished); [19]- Hand; [20]- end/start; [21]- Hands and Head position; [22]- only ASL fingerspelling sequences.

Dataset information and related papers

  1. DGS Kinect 40 - German Sign Language
    1. Sign Language Recognition using Sub-Units, 2012, Cooper et al.
    2. Sign Language Recognition using Sequential Pattern Trees 2012, Ong et al.
    3. Sign Spotting using Hierarchical Sequential Patterns with Temporal Intervals 2014, Ong et al.
  2. RWTH-PHOENIX v1 - German Sign Language RWTH-PHOENIX v2
    1. Dataset paper 2012, Forster et al.
    2. Dataset extensions paper 2014, Forster et al
    3. Continuous sign language recognition: Towards large vocabulary statistical recognition systems handling multiple signers 2015, Koller et al.
    4. May the force be with you: Force-aligned signwriting for automatic subunit annotation of corpora 2013, Koller et al.
    5. Deep Sign: Hybrid CNN-HMM for Continuous Sign Language Recognition
  3. SIGNUM - German Sign Language
    1. Continuous sign language recognition: Towards large vocabulary statistical recognition systems handling multiple signers 2015, Koller et al.
  4. Greek Sign Language (no website)
    1. Sign Language Recognition using Sub-Units, 2012, Cooper et al.
    2. Sign Language Recognition using Sequential Pattern Trees 2012, Ong et al.
    3. Sign Spotting using Hierarchical Sequential Patterns with Temporal Intervals 2014, Ong et al.
  5. Boston ASLLVD - American Sign Language
    1. Exploiting Phonological Constraints for Handshape Inference in ASL Video 2011, Thangali et al.
    2. A New Framework for Sign Language Recognition based on 3D Handshape Identification and Linguistic Modeling 2014 - Dilsizian - 84% accuracy
  6. PSL Kinect 30 - Polish Sign Language
    1. Polish sign language words recognition with Kinect 2013, Oszust et al.
    2. Some Approaches to Recognition of Sign Language Dynamic Expressions with Kinect 2014, Oszust et al.
    3. Recognition of Hand Gestures Observed by Depth Cameras 2015, Kapuscinski et al.
  7. PSL ToF 84 - Polish Sign Language
    1. Polish sign language words recognition with Kinect 2013,Oszust et al.
    2. Recognition of Hand Gestures Observed by Depth Cameras 2015, Kapuscinski et al.
  8. PSL 101 - Polish Sign Language (no website)
    1. Modelling and Recognition of Signed Expressions Using Subunits Obtained by Data–Driven Approach 2012, Oszust et al.
  9. LSA64 Argentinian Sign Language
    1. LSA64: an Argentinian Sign Language Dataset
    2. Sign Languague Recognition Without Frame-Sequencing Constraints: A Proof of Concept on the Argentinian Sign Language
    3. Dynamic Gesture Recognition and its Application to Sign Language 2017, Ronchetti
    4. SIGN LANGUAGE RECOGNITION BASED ON HAND AND BODY SKELETAL DATA 2017,Konstantinidis et al.
    5. Real-Time Sign Language Gesture (Word) Recognition from Video Sequences Using CNN and RNN 2018, Masood et al.
  10. Turkish sign language dataset
  11. MSR Gesture 3D - ASLDownload site
    1. Action Recognition from Depth Sequences Using Weighted Fusion of 2D and 3D Auto-Correlation of Gradients Features 2016, Chen et al
  12. DEVISIGN G
  13. DEVISIGN D
  14. DEVISIGN L
  15. IIITA -ROBITA Indian Sign Language Gesture Database
    1. Recognizing & Interpreting Indian Sign Language Gesture for Human Robot Interaction 2010, Nandy et al.
    2. Recognition of Isolated Indian Sign Language gesture in Real Time 2010, Nandy et al.
  16. Purdue ASL Dataset
  17. CUNY ASL Dataset for Animation
    1. Collecting and evaluating the CUNY ASL corpus for research on American Sign Language animation
  18. SignsWorld Atlas; a benchmark Arabic Sign Language database

Datasets Handshape features (Handshape/hand posture datasets) but not all are for sign language

id Name Co Clas Sub Samples Data Type Availability
1 ASL Fingerspelling A USA 24 5 131000 2.1gb images (depth+rgb) Free download
2 ASL Fingerspelling B USA 24 9 317mb images (depth) Free download
3 LSA16 handshapes Arg 16 10 800 7mb images (rgb) Free download
4 PSL Fingerspelling ToF Pol 16 3 960 ~290mb 3D point cloud Free download
5 ISL Iri [23] 6 [24] 170mb segmented images Free download
6 RWTH-PHOENIX-Weather Handshapes Ger 60 [25] + 17gb Hand Images (rgb) Free download
7 Japanese Fingerspelling Dataset Jap 41 10 8055 4.5mb [26] Free download
8 NUS hand posture dataset I Sin 10 ? 240 3mb images(rgb),160x120 Free download
9 NUS hand posture dataset II Sin 10 40 2000 73mb images(rgb)160x120 Free download
10 CIARP - 10 ? 6000 11mb images(rgb)38x38 Free download
11 RTWH Fingerspelling dataset Ger
12 Indian Kinect Ind 40 18 5041 2gb [27] Free download
13 [ArASL] Ara 32 ? 54,049 64mb images(rgb) Free download
14 ChicagoFSWild USA [2] 160 ? images(rgb) Free download
15 ChicagoFSWild+ USA

[ArASL] - Arabic Alphabets Sign Language Dataset; [2] - multiple types; [23]- 23 static + 3 dynamic; [24]- 58114 frames/468 videos; [25]- 3359 labelled + 17gb unlabeled [26]- segmented images (rgb), 32x32 [27]- images (rgb+depth) 640x480

Dataset information and related papers

  1. ASL Fingerspelling
    1. Spelling It Out: Real-Time ASL Fingerspelling Recognition. 2011, Pugeault et al.
    2. Recognition of Hand Gestures Observed by Depth Cameras. 2015, Kapuscinski et al.
  2. PSL Fingerspelling ToF
    1. Recognition of Hand Gestures Observed by Depth Cameras. 2015, Kapuscinski et al.
  3. LSA16 handshapes
    1. Handshape recognition for Argentinian Sign Language using ProbSom. 2016, Ronchetti et al.
    2. A Study of Convolutional Architectures for Handshape Recognition applied to Sign Language 2017, Quiroga et al.
  4. ISL Irish Sign Language Letters.
    1. A Dataset for Irish Sign Language Recognition 2017, Oliveira et al.
    2. A comparison between end-to-end approaches and feature extraction based approaches for Sign Language recognition 2017, Oliveira et al.
  5. RWTH-PHOENIX-Weather 2014 MS Handshapes
    1. Deep Hand: How to Train a CNN on 1 Million Hand Images When Your Data Is Continuous and Weakly Labelled  2017, Koller et al.
  6. Japanese Sign Language Dataset
    1. Recognition of JSL Finger Spelling Using Convolutional Neural Networks 2017, Hosoe, Sako and Kwolek
    2. Learning Siamese Features for Finger Spelling Recognition 2017, Sako and Kwolek
  7. NUS hand posture dataset I
    1. Hand posture and face recognition using a Fuzzy-Rough Approach 2010, Pramod Kumar P, Prahlad Vadakkepat, and Loh Ai Poh
    2. Hand Posture Recognition Using Convolutional Neural Network
  8. NUS hand posture dataset II
    1. Attention Based Detection and Recognition of Hand Postures Against Complex Backgrounds 2013, Pisharady et al
  9. CIARP 2017
    1. Hand Posture Recognition Using Convolutional Neural Network
  10. RTWH Fingerspelling dataset
  11. Modeling Image Variability in Appearance-Based Gesture Recognition. In ECCV Workshop on Statistical Methods in Multi-Image and Video Processing
  12. Indian Kinect github
  13. Nearest neighbour classification of Indian sign language gestures using kinect camera 2016, Ansari and Harit
  14. Arabic Alphabets Sign Language Dataset (ArASL)
  15. Arabic Alphabet and Numbers Sign Language Recognition
  16. ChicagoFSWild
    1. American Sign Language fingerspelling recognition in the wild
    2. Fingerspelling recognition in the wild with iterative visual attention
  17. ChicagoFSWild+
    1. Fingerspelling recognition in the wild with iterative visual attention

Continuous hand pose

  1. NYU Hand pose dataset
    1. Real-Time Continuous Pose Recovery of Human Hands Using Convolutional Networks

Datasets of facial features

  1. QLIBRAS
    1. QLIBRAS: A novel database for grammatical facial expressions in Brazilian Sign Language
  2. WIDER FACE: A Face Detection Benchmark
    1. Fingerspelling recognition in the wild with iterative visual attention

Datasets of lip reading features

  1. GRID corpus - Lip
    1. LipNet: End-to-End Sentence-level Lipreading
    2. Dynamic Stream Weighting for Turbo-Decoding-Based Audiovisual ASR
  2. AVICAR - Lip
    1. Classification and Feature Extraction by Simplexization
  3. AVLetter
    1. Temporal Multimodal Learning in Audiovisual Speech Recognition
  4. CUAVE
    1. CUAVE: A new audio-visual database for multimodal human-computer interface research
  5. OuluVS1 (no website)
    1. Lip Reading in the Wild
  6. OuluVS2
    1. Out of time: automated lip sync in the wild
  7. BBC TV
    1. Lip Reading in the Wild

The table from the paper - LipNet: End-to-End Sentence-level Lipreading

Method Dataset Size Output Accuracy
Fu et al. (2008) AVICAR 851 Digits 37.9%
Hu et al. (2016) AVLetter 78 Alphabet 64.6%
Papandreou et al. (2009) CUAVE 1800 Digits 83.0%
Chung & Zisserman (2016a) OuluVS1 200 Phrases 91.4%
Chung & Zisserman (2016b) OuluVS2 520 Phrases 94.1%
Chung & Zisserman (2016a) BBC TV > 400000 Words 65.4%
Gergen et al. (2016) GRID 29700 Words 86.4%
LipNet GRID 28775 Sentences 95.2%

Other info Kevin Murphy mantains a similar list for Action Recognition Datasets. Other similar websites with sign language dataset compilations are:

Papers that cite datasets that are unavailable:

  1. LIBRAS hand movement
    1. Hand Movement Recognition for Brazilian Sign Language: A Study Using Distance-Based Neural Networks.

References https://facundoq.github.io/unlp/sign_language_datasets/index.html

About

These datasets are used for machine-learning research