The 20BN-jester Dataset V1


Introduction

The 20BN-JESTER dataset is a large collection of labeled video clips that show humans performing pre-definded hand gestures in front of a laptop camera or webcam. The dataset was created by a large number of crowd workers. It allows for training robust machine learning models to recognize human hand gestures. It is available free of charge for academic research. Commercial licenses are available upon request.

A paper with supplementary material can be found here.

Sliding Two Fingers Down
Swiping Left
Thumb Up

Data format

The video data is provided as one large TGZ archive, split into parts of 1 GB max. The total download size is 22.8 GB. The archive contains directories numbered from 1 to 148092. Each directory corresponds to one video and contains JPG images with height 100px and variable width. The JPG images were extracted from the orginal videos at 12 frames per seconds. The filenames of the JPGs start at 00001.jpg. The number of JPGs varies as the length of the original videos varies.


20BN-JESTER-DATASET
Total number of videos
148,092
Training Set
118,562
Validation Set
14,787
Test Set (w/o labels)
14,743
Labels
27
Classes
12,416
Doing other things
5,460
Thumb Down
5,457
Thumb Up
5,444
Drumming Fingers
5,434
Pushing Hand Away
5,413
Stop Sign
5,410
Sliding Two Fingers Down
5,379
Pulling Hand In
5,379
Zooming Out With Two Fingers
5,358
Pushing Two Fingers Away
5,355
Zooming In With Two Fingers
5,345
Sliding Two Fingers Left
5,344
No gesture
5,330
Zooming Out With Full Hand
5,315
Pulling Two Fingers In
5,314
Shaking Hand
5,307
Zooming In With Full Hand
5,303
Swiping Down
5,262
Sliding Two Fingers Up
5,244
Sliding Two Fingers Right
5,240
Swiping Up
5,165
Rolling Hand Forward
5,160
Swiping Left
5,066
Swiping Right
5,031
Rolling Hand Backward
4,181
Turning Hand Counterclockwise
3,980
Turning Hand Clockwise

Available Licenses

Twenty Billion Neurons offers our Crowd Acting™ video dataset collections in three different license types depending on the organization you belong to and the intended use for the data.

Free Academic License

Academic institutions and non-profit organizations only.

Proceed

Corporate Research License

Perform research and evaluations in a corporate research lab or for-profit organization.

Proceed

Full Commercial License

Build the next level commercial A.I. products and services.

Proceed

Leaderboard

If you have been successful in creating a classification model based on the training set and it performs well on the validation set, we encourage you to run your model on the test set (which is published without any class labels, as you might have noticed). Please prepare a .csv file with the video's id in the first column and your predicted class label (as a string matching the wording used in the training and validation sets). As a separator, please use a semicolon. You can then upload your .csv file here (user login required) to be ranked in the leaderboard and to benchmark your approach against that of other machine learners. We are looking forward to your submission.

Rank
Name
Approach
Score
1
TmallGenie GestureWYS
5 months ago

WYS

97.37%
2
Iflytek
over 1 year ago

MobileNet+NL+SlowFast

97.26%
3
ysge
7 months ago

ysge

97.23%
4
BOE_IOT_AIBD
almost 2 years ago

Fusion_TSN_LSTM

97.09%
5
Anonymous
over 1 year ago

CVPR2020Submission

97.09%
6
Huawei Noah's Ark Toronto Laboratory Team
about 2 years ago

RFEEN, 20 Crops

97.06%
7
zhang yuan
over 1 year ago

MobileNet_v2_NL_16sample

97.04%
8
Anonymous
about 1 year ago

12F rgb

96.99%
9
Anonymous
almost 2 years ago

rgb 12F

96.95%
10
Gaurav Kumar Singh
over 2 years ago

Ford's Gesture Recognition System

96.77%
11
Anonim
almost 3 years ago

96.74%
12
NLPR-CASIA-LSHI
over 3 years ago

L. Shi, Y. Zhang, C. Jian, and L. Hanqing, "Gesture Recognition using Spatiotemporal Deformable Convolutional Representation" in IEEE International Conference on Image Processing (ICIP), 2019.

96.6%
13
Anonymous
over 2 years ago

96.56%
14
fengye
5 months ago

Deep fusion model

96.41%
15
Okan Köpüklü
over 3 years ago

Motion Fused Frames (MFFs)
Code: https://github.com/okankop/MFF-pytorch
Article: https://arxiv.org/pdf/1804.07187.pdf
Contact:okankopuklu@gmail.com

96.28%
16
Mohamad ALjazaery @Midea
about 3 years ago

Spatiotemporal Two Streams network

96.28%
17
Anonymous
about 3 years ago

3D CNN Architecture

96.24%
18
Anonymous
over 3 years ago

Motion Feature Network (MFNet)

96.22%
19
Villa
12 months ago

96.16%
20
guandai
5 months ago

96.11%
21
Anonymous
over 2 years ago

RNP

95.96%
22
Jingyao Wang
over 2 years ago

95.96%
23
jiaming huang
over 1 year ago

95.94%
24
Anonymous
over 2 years ago

ResNext 101

95.87%
25
anonymous
over 2 years ago

SSNet RGB resnet

95.79%
26
Anonymous
over 1 year ago

95.79%
27
煌彬 吴
over 1 year ago

95.74%
28
Anonymous
over 1 year ago

95.74%
29
Wenjin Zhang, Jiacun Wang
over 1 year ago

Short-Term Sampling Neural Networks (9 groups)

95.73%
30
Anonymous
over 2 years ago

TVB

95.71%
31
Ke Yang (NUDT_PDL)
over 3 years ago

Temporal Pyramid Relation Network for Video-Based Gesture Recognition,2018 25th IEEE International Conference on Image Processing (ICIP)

95.34%
32
Anonymous
over 3 years ago

DIN

95.31%
33
Anonymous
over 2 years ago

95.14%
34
Guangming Zhu
over 3 years ago

95.01%
35
minivision_art
over 1 year ago

94.97%
36
Test
about 2 years ago

TRN - 8 segments

94.95%
37
lilh
over 1 year ago

94.93%
38
Anonymous
over 1 year ago

94.88%
39
Wenjin Zhang
over 1 year ago

Short-Term Sampling Neural Networks

94.88%
40
SJ
over 3 years ago

94.87%
41
Roy Amante Salvador
over 2 years ago

3D CNN - Multi time scale evaluation

94.85%
42
Anonymous
over 2 years ago

8frames rgb

94.81%
43
Anonymous
over 3 years ago

TRN (CVPR'18 submission)

94.78%
44
Thomas Friedel
over 3 years ago

94.74%
45
ALAB
almost 3 years ago

TRN + BNInception

94.5%
46
Anonymous
about 2 years ago

Anonymous

94.49%
47
Anonymous
over 2 years ago

final test
label+string

94.47%
48
shi huiying
8 months ago

94.36%
49
Roy Salvador
over 2 years ago

3D CNN for transfer learning

94.26%
50
Eren Gölge
almost 4 years ago

Besnet

94.23%
51
Wu Jie @ DLUT-SIE
over 2 years ago

3D_GesNet

93.99%
52
Wu Jie @ DLUT-SIE
over 2 years ago

3D-GesNet(only rgb)

93.99%
53
Guillaume Berger
almost 4 years ago

93.87%
54
XM
over 2 years ago

ECO

93.82%
55
ming chen
over 1 year ago

2D and 3D fused network

93.8%
56
Anonymous
over 2 years ago

TRN-E

93.58%
57
Zubair
almost 2 years ago

Inceptionv2 - TRN16

93.55%
58
xxx
11 months ago

Test 1

93.47%
59
long
11 months ago

test

93.47%
60
666 212
11 months ago

test rgb 16f

93.47%
61
Francesco Dalla Serra
almost 3 years ago

One Stream Modified-I3D

93.41%
62
DIVA
11 months ago

TAN_ep2

93.39%
63
199_3D
over 1 year ago

199_3D

93.12%
64
l lln
over 1 year ago

p3d_199

93.12%
65
Baptist
over 3 years ago

93.11%
66
Fábio Baldissera
about 2 years ago

RT C3D - 16 Frames
Code: https://github.com/fabiopk

92.65%
67
INF
11 months ago

TPP sub1

92.52%
68
RyMult
almost 2 years ago

91.61%
69
rml
almost 2 years ago

91.61%
70
Prasad Pai
over 2 years ago

91.45%
71
Zhaoli Deng
about 1 year ago

91.39%
72
Jerry Tom
about 1 year ago

经典双赢

91.39%
73
wuxia tu
over 1 year ago

prune_net

91.22%
74
63_3D
over 1 year ago

63_3D

91.19%
75
tu tututu
over 1 year ago

result_3Dconv_63

91.19%
76
Yuriy
5 months ago

mobilenetv2_1.00_160, Dense, Dense

90.84%
77
Oskar Holmberg, Filip Granqvist
about 3 years ago

90.52%
78
anonymous
about 3 years ago

Modified C3D

89.26%
79
qi yuan
over 2 years ago

87.94%
80
Haibing Huang
over 2 years ago

CNN+LSTM

86.31%
81
Anonymous
over 1 year ago

3D CNN

86.15%
82
Zhiyuan Ma
over 1 year ago

86.07%
83
Arnaud Steinmetz
over 2 years ago

3D ResNet 101

85.99%
84
John Emmons
over 3 years ago

VideoLSTM

85.86%
85
Olivier Valery
almost 3 years ago

3D convolutional neural network

85.49%
86
qiu feng
over 1 year ago

Result_63

85.31%
87
Damien MENIGAUX
over 3 years ago

ConvLSTM

82.76%
88
Raghul Krishna
over 1 year ago

82.53%
89
Joanna Materzyńska
about 4 years ago

Twenty Billion Neuron's Jester System

82.34%
90
lbjbx
about 2 years ago

3d+resnet18

81.55%
91
Jakub B.
6 months ago

ColumnConv3D

78.74%
92
Wei-Cheng Lin
almost 2 years ago

77.89%
93
Liam Schoneveld
about 2 years ago

Basic finetune MobileNetV2 (pretrained imagenet) + LSTM output

74.4%
94
GY@CAD
almost 2 years ago

interframe-difference LSTM

73.22%
95
yuzhe zhou
almost 2 years ago

Mobilenetv3-LSTM

71.72%
96
ke yang(BUPT)
about 3 years ago

3D ResNet

68.13%
97
John Waithaka
8 months ago

CNN-LSTM

63.0%
98
Victor
almost 3 years ago

3D-ResNet101 trained on Kinetics

59.01%
99
CS2R_Muneer
over 1 year ago

57.04%
100
PortlyBoi
8 months ago

3 Temporal Stream Network + ConvRNN

48.23%
101
FeiHu Jiang
almost 2 years ago

2D+3D

44.04%
102
James Zhang
5 months ago

2-Frame Difference with Otsu+One Stream I3D

39.37%
103
Yu Zhu
almost 4 years ago

10.52%
104
Ming
about 3 years ago

Test

4.0%
105
Anonymous
over 2 years ago

3.97%
106
Anonymous
over 2 years ago

rgb_only
test label (from 0 to 26)

0.0%
107
Anonymous
over 2 years ago

ResNext 101

0.0%
108
Fabio B.
about 2 years ago

[test_run] 3D RGB 16F

0.0%
109
ozgur
about 2 years ago

0.0%



Feedback