JUHE API Marketplace
Back to datasets
Dataset assetOpen Source CommunityMachine LearningSign Language Recognition

WLASL, DGS Kinect 40, RWTH-PHOENIX-Weather, SIGNUM, GSL 20, Boston ASL LVD, PSL Kinect 30, PSL ToF 84, PSL 101, LSA64, BosphorusSign, MSR Gesture 3D, DEVISIGN-G, DEVISIGN-D, DEVISIGN-L, IIITA -ROBITA, Purdue ASL, CUNY ASL, SignsWorld Atlas

These datasets are used for machine‑learning research and encompass sign language data from various countries and languages, including video and depth data, for sign language recognition and understanding studies.

Source
github
Created
Dec 10, 2019
Updated
May 24, 2024
Signals
329 views
Availability
Linked source ready
Overview

Dataset description and usage context

Dataset Overview

Main Dataset List

1. DGS Kinect 40

  • Country: Germany
  • Number of Classes: 40
  • Number of Participants: 15
  • Number of Samples: 3000
  • Data Type: Video
  • Language Level: Words
  • Availability: Public
  • Data Storage: Hard Disk

2. RWTH-PHOENIX-Weather

  • Country: Germany
  • Number of Classes: 1200
  • Number of Participants: 9
  • Number of Samples: 45760
  • Data Size: 52GB
  • Data Type: Video
  • Language Level: Sentences
  • Availability: Public
  • Data Storage: Hard Disk

3. SIGNUM

  • Country: Germany
  • Number of Classes: 450
  • Number of Participants: 25
  • Number of Samples: 33210
  • Data Size: 920GB
  • Data Type: Video
  • Availability: Public
  • Data Storage: None

4. GSL 20

  • Country: Greece
  • Number of Classes: 20
  • Number of Participants: 6
  • Number of Samples: ~840
  • Data Type: Video
  • Availability: Contact author
  • Data Storage: Hard Disk

5. Boston ASL LVD

  • Country: United States
  • Number of Classes: 3300+
  • Number of Participants: 6
  • Number of Samples: 9800
  • Data Type: Video
  • Language Level: Words
  • Availability: Public
  • Data Storage: None

6. PSL Kinect 30

  • Country: Poland
  • Number of Classes: 30
  • Number of Participants: 1
  • Number of Samples: 300
  • Data Size: ~1.2GB
  • Data Type: Video
  • Language Level: Words
  • Availability: Public
  • Data Storage: Hard Disk

7. PSL ToF 84

  • Country: Poland
  • Number of Classes: 84
  • Number of Participants: 1
  • Number of Samples: 1680
  • Data Size: ~33GB
  • Data Type: Video
  • Language Level: Words
  • Availability: Public
  • Data Storage: None

8. PSL 101

  • Country: Poland
  • Number of Classes: ?
  • Number of Participants: ?
  • Number of Samples: ?
  • Data Type: ?
  • Availability: Contact author
  • Data Storage: None

9. LSA64

  • Country: Argentina
  • Number of Classes: 64
  • Number of Participants: 10
  • Number of Samples: 3200
  • Data Size: 20GB
  • Data Type: Video (RGB)
  • Availability: Public
  • Data Storage: Hard Disk

10. BosphorusSign

  • Country: Turkey
  • Number of Classes: None
  • Number of Participants: None
  • Number of Samples: None
  • Data Type: None
  • Availability: Unavailable
  • Data Storage: None

11. MSR Gesture 3D

  • Country: United States
  • Number of Classes: 12
  • Number of Participants: 10
  • Number of Samples: 336
  • Data Size: 28MB
  • Data Type: Video (Depth)
  • Availability: Public
  • Data Storage: None

12. DEVISIGN-G

  • Country: China
  • Number of Classes: 36
  • Number of Participants: 8
  • Number of Samples: 432
  • Data Type: Video (RGB)
  • Availability: Contact author
  • Data Storage: None

13. DEVISIGN-D

  • Country: China
  • Number of Classes: 500
  • Number of Participants: 8
  • Number of Samples: 6000
  • Data Type: Video (RGB)
  • Availability: Contact author
  • Data Storage: None

14. DEVISIGN-L

  • Country: China
  • Number of Classes: 2000
  • Number of Participants: 8
  • Number of Samples: 24000
  • Data Type: Video (RGB)
  • Availability: Contact author
  • Data Storage: None

15. IIITA -ROBITA

  • Country: India
  • Number of Classes: 23
  • Number of Participants: ?
  • Number of Samples: None
  • Data Size: 284MB
  • Data Type: Video (RGB)
  • Availability: Contact author
  • Data Storage: None

16. Purdue ASL

  • Country: United States
  • Number of Classes: ?
  • Number of Participants: 14
  • Number of Samples: ?
  • Data Type: Video
  • Language Level: Words/Sentences
  • Availability: Contact author
  • Data Storage: None

17. CUNY ASL

  • Country: United States
  • Number of Classes: ?
  • Number of Participants: 8
  • Number of Samples: ~33000
  • Data Type: Video (RGB)
  • Language Level: Sentences
  • Availability: Unknown
  • Data Storage: None

18. SignsWorld Atlas

  • Country: Arab countries
  • Number of Classes: None
  • Number of Participants: 10
  • Number of Samples: None
  • Data Type: Video
  • Language Level: Words, Sentences, Handshapes
  • Availability: Unknown
  • Data Storage: None

Gesture Feature Datasets

1. ASL Fingerspelling A

  • Country: United States
  • Number of Classes: 24
  • Number of Participants: 5
  • Number of Samples: 131000
  • Data Size: 2.1GB
  • Data Type: Image (Depth+RGB)
  • Availability: Free download

2. ASL Fingerspelling B

  • Country: United States
  • Number of Classes: 24
  • Number of Participants: 9
  • Number of Samples: None
  • Data Size: 317MB
  • Data Type: Image (Depth)
  • Availability: Free download

3. LSA16 handshapes

  • Country: Argentina
  • Number of Classes: 16
  • Number of Participants: 10
  • Number of Samples: 800
  • Data Size: 7MB
  • Data Type: Image (RGB)
  • Availability: Free download

4. PSL Fingerspelling ToF

  • Country: Poland
  • Number of Classes: 16
  • Number of Participants: 3
  • Number of Samples: 960
  • Data Size: ~290MB
  • Data Type: 3D point cloud
  • Availability: Free download

5. ISL

  • Country: Ireland
  • Number of Classes: 23
  • Number of Participants: 6
  • Number of Samples: 58114 frames / 468 videos
  • Data Size: 170MB
  • Data Type: Segmentation images
  • Availability: Free download

6. RWTH-PHOENIX-Weather Handshapes

  • Country: Germany
  • Number of Classes: 60
  • Number of Participants: None
  • Number of Samples: 3359 labeled + 17GB unlabeled
  • Data Size: Over 17GB
  • Data Type: Hand images (RGB)
  • Availability: Free download

7. Japanese Fingerspelling Dataset

  • Country: Japan
  • Number of Classes: 41
  • Number of Participants: 10
  • Number of Samples: 8055
  • Data Size: 4.5MB
  • Data Type: Segmentation images (RGB), 32×32
  • Availability: Free download

8. NUS hand posture dataset I

  • Country: Singapore
  • Number of Classes: 10
  • Number of Participants: ?
  • Number of Samples: 240
  • Data Size: 3MB
  • Data Type: Image (RGB), 160×120
  • Availability: Free download

9. NUS hand posture dataset II

  • Country: Singapore
  • Number of Classes: 10
  • Number of Participants: 40
  • Number of Samples: 2000
  • Data Size: 73MB
  • Data Type: Image (RGB), 160×120
  • Availability: Free download

10. CIARP 2017

  • Country: None
  • Number of Classes: 10
  • Number of Participants: ?
  • Number of Samples: 6000
  • Data Size: 11MB
  • Data Type: Image (RGB), 38×38
  • Availability: Free download

11. RTWH Fingerspelling dataset

  • Country: Germany
  • Number of Classes: None
  • Number of Participants: None
  • Number of Samples: None
  • Data Type: None
  • Availability: None

12. Indian Kinect

  • Country: India
  • Number of Classes: 40
  • Number of Participants: 18
  • Number of Samples: 5041
  • Data Size: 2GB
  • Data Type: Image (RGB+Depth) 640×480
  • Availability: Free download

13. ArASL

  • Country: Arab countries
  • Number of Classes: 32
  • Number of Participants: ?
  • Number of Samples: 54049
  • Data Size: 64MB
  • Data Type: Image (RGB)
  • Availability: Free download

14. ChicagoFSWild

  • Country: United States
  • Number of Classes: 2
  • Number of Participants: 160
  • Number of Samples: None
  • Data Type: Image (RGB)
  • Availability: Free download

15. ChicagoFSWild+

  • Country: United States
  • Number of Classes: None
  • Number of Participants: None
  • Number of Samples: None
  • Data Type: None
  • Availability: None

Continuous Hand Pose Datasets

1. NYU Hand pose dataset

  • Data Type: None
  • Availability: None
  • Data Storage: None

Facial Feature Datasets

1. QLIBRAS

  • Data Type: None
  • Availability: None
  • Data Storage: None

2. WIDER FACE: A Face Detection Benchmark

  • Data Type: None
  • Availability: None
  • Data Storage: None

Lip Reading Feature Datasets

1. GRID corpus - Lip

  • Data Type: None
  • Availability: None
  • Data Storage: None

2. AVICAR - Lip

  • Data Type: None
  • Availability: None
  • Data Storage: None

3. AVLetter

  • Data Type: None
  • Availability: None
  • Data Storage: None

4. CUAVE

  • Data Type: None
  • Availability: None
  • Data Storage: None

5. OuluVS1

  • Data Type: None
  • Availability: None
  • Data Storage: None

6. OuluVS2

  • Data Type: None
  • Availability: None
  • Data Storage: None
Need downstream help?

Pair the dataset with AI analysis and content workflows.

Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.

Explore AI studio