-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathindex.html
234 lines (189 loc) · 9.04 KB
/
index.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width,initial-scale=1">
<title> Human Foot Keypoint Dataset </title>
<link href="css/normalize.css" rel="stylesheet">
<link type="text/css" rel="stylesheet" href="css/prism.css"/>
<!-- Latest compiled and minified CSS -->
<link rel="stylesheet" href="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.6/css/bootstrap.min.css" integrity="sha384-1q8mTJOASx8j1Au+a5WDVnPi2lkFfwwEAa8hDDdjZlpLegxhjVME1fgjWPGmkzs7" crossorigin="anonymous">
<link href="css/main.css" rel="stylesheet">
<!-- Optional theme -->
<link rel="stylesheet" href="css/bootstrap.min.css">
<script type="text/javascript" src="http://ajax.googleapis.com/ajax/libs/jquery/1.9.1/jquery.min.js"></script>
<!-- Latest compiled and minified JavaScript -->
<script src="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.6/js/bootstrap.min.js" integrity="sha384-0mSbJDEHialfmuBBQP6A4Qrprq5OVfW37PRR3j5ELqxss1yVqOtnepnHVP9aJ7xS" crossorigin="anonymous"></script>
<style>
body {
padding-top: 60px; /* 60px to make the container go all the way to the bottom of the topbar */
}
</style>
<link href="css/bootstrap-responsive.css" rel="stylesheet">
<!-- HTML5 shim, for IE6-8 support of HTML5 elements -->
<!--[if lt IE 9]>
<script src="../assets/js/html5shiv.js"></script>
<![endif]-->
<link href="css/myStyle.css " rel="stylesheet">
<link href="css/thumbnail.css" rel="stylesheet">
<style type="text/css" media="screen">
table td.highlighted{background-color:#87CEFA;}
</style>
<style>
h2
{
text-align: left;
}
p.note
{
font-family: Arial Black;
font-size: 14px;
}
ul.note
{
list-style-type: none;
padding-left: 0px;
}
li.note {
font-family: Verdana;
position: relative;
padding-left: 7px;
font-size: 13px;
}
li.note:before {
content: "\e080";
font-family: 'Glyphicons Halflings';
font-size: 9px;
position: relative;
margin-right: 3px;
top: 3px;
color: #ccc;
}
</style>
</head>
<body>
<div class="container">
<h1> <center> Human Foot Keypoint Dataset</center> </h1>
<p align = "center"><font size = "4"> <a href="https://people.eecs.berkeley.edu/~zhecao/">Zhe Cao</a>, <a href="https://www.gineshidalgo.com/">Gines Hidalgo</a>, <a href="http://www.cs.cmu.edu/~tsimon/">Tomas Simon</a>, <a href="https://scholar.google.com/citations?user=sFQD3k4AAAAJ&hl=en">Shih-En Wei</a>, <a href="http://www.cs.cmu.edu/~yaser/">Yaser Sheikh</a></font></p>
<p align = "center"><font size = "4"> Carnegie Mellon University </font></p>
<div id="teaser">
<img src="fig_feet.png" width="100%" alt="Foot examples">
</div>
<h2>Description</h2>
<b>Quoting <a href="https://arxiv.org/abs/1812.08008"><i>OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields</i></a></b>:
<br>
<p><i>
Existing human pose datasets contain limited body part types. The MPII dataset annotates ankles, knees, hips, shoulders, elbows, wrists, necks, torsos, and head tops, while COCO also includes some facial keypoints. For both of these datasets, foot annotations are limited to ankle position only. However, graphics applications such as avatar retargeting or 3D human shape reconstruction require foot keypoints such as big toe and heel. Without foot information, these approaches suffer from problems such as the candy wrapper effect, floor penetration, and foot skate. To address these issues, <b>a small subset of foot instances out of the COCO dataset is labeled using the Clickworker platform. It is split up with 14K annotations from the COCO training set and 545 from the validation set. A total of 6 foot keypoints are labeled.</b> We consider the 3D coordinate of the foot keypoints rather than the surface position. For instance, for the exact toe positions, we label the area between the connection of the nail and skin, and also take depth into consideration by labeling the center of the toe rather than the surface.
</i></p>
<br>
The foot keypoint <b>ordering</b> in the annotation file is as follows:
<br>
<table border="0">
<tr>
<td align="center"><img src="foot_keypoints.png" alt="photo" width="110" vspace="5" hspace="25" /></td>
<td>
<ol type="1" start="0">
<br>
<li>Left big toe.</li>
<li>Left small toe.</li>
<li>Left heel.</li>
<li>Right big toe.</li>
<li>Right small toe.</li>
<li>Right heel.</li>
</ol>
</td>
</tr>
</table>
<!-- <center>
<img src="foot_keypoints.png" width="150" alt="Foot examples">
</center> -->
<br>
<h2>Download</h2>
<div id="stats-table">
<table class="table table-striped" align="left">
<tr>
<td>
<h5><b>Foot Keypoint Annotations (Training: ~13.5k annotations, Validation: ~0.5k annotations)</b></h5>
</td>
</tr>
<tr>
<td>
<!-- <a href="http://domedb.perception.cs.cmu.edu/panopticDB/hands/hand_labels.zip">Download (588MB)</a> -->
Download the <a href="http://posefs1.perception.cs.cmu.edu/OpenPose/datasets/foot/person_keypoints_train2017_foot_v1.zip"><b>train2017_foot_v1.zip</b></a> JSON zip file.
</td>
</tr>
<tr>
<td>
Download the <a href="http://posefs1.perception.cs.cmu.edu/OpenPose/datasets/foot/person_keypoints_val2017_foot_v1.zip"><b>val2017_foot_v1.zip</b></a> JSON zip file.
</td>
</tr>
<tr>
<td>
Download the image dataset from the <a href="http://cocodataset.org/#download" target="_blank"><b>original body-only dataset website</b></a> (in particular <a href="http://images.cocodataset.org/zips/train2017.zip"><b>2017 Train images [118K/18GB]</b></a> and <a href="http://images.cocodataset.org/zips/val2017.zip"><b>2017 Val images [5K/1GB]</b></a>)
</td>
</tr>
<tr>
<td>
Trained foot detector models and testing code released in <a href="https://github.com/CMU-Perceptual-Computing-Lab/openpose" target="_blank"><b>OpenPose</b></a>
</td>
</tr>
</table>
</div>
<p>
The JSON files follow the same standard format than that of <a href="http://images.cocodataset.org/annotations/annotations_trainval2017.zip"><b>2017 Train/Val annotations [241MB]</b></a> (from the <a href="http://cocodataset.org/#download" target="_blank"><b>original body-only dataset website</b></a>).
</p>
<br>
<h2>Training Code</h2>
<p>
For training code, check <a href="https://github.com/CMU-Perceptual-Computing-Lab/openpose_train" target="_blank"><b>https://github.com/CMU-Perceptual-Computing-Lab/openpose_train</b></a>.
</p>
<br>
<h2>Results</h2>
<center>
<a href="https://youtu.be/Cu7g2Ldm-WM" target="_blank"> <img src="results_flashmob.gif"
width="640"
alt="Foot examples">
<!-- Full video here: <b>here</b> -->
</a>.
<!-- width="560" height="315" (youtube recommneded)-->
<!-- allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" -->
<!-- width="1024" height="576" -->
<!-- <iframe
width="960" height="540"
src="https://www.youtube.com/embed/XPBw1IvixWg" frameborder="0"
allowfullscreen></iframe> -->
<!-- <iframe width="1024" height="450" src="https://www.youtube.com/embed/Lajt6vS_dSM" frameborder="0" allowfullscreen></iframe> -->
<!-- <tr>
<td>
<img src="fig_feet.png"
width="1024"
alt="Foot examples">
height="200"
</td>
</tr> -->
</center>
<br>
<h2>Citation</h2>
<h5>Please cite the following paper (<a href="https://arxiv.org/abs/1812.08008">arXiv link</a>) in your publications if the dataset helps your research.</h5>
@inproceedings{cao2018openpose,
<br>
author = {Zhe Cao and Gines Hidalgo and Tomas Simon and Shih-En Wei and Yaser Sheikh},
<br>
booktitle = {arXiv preprint arXiv:1812.08008},
<br>
title = {Open{P}ose: realtime multi-person 2{D} pose estimation using {P}art {A}ffinity {F}ields},
<!-- title = {Open{P}ose: Realtime {M}ulti-{P}erson 2{D} {P}ose {E}stimation using {P}art {A}ffinity {F}ields}, -->
<br>
year = {2018}
<br>
}
<br>
<br>
<h2>License</h2>
<p>The software and Foot Keypoint Annotations entitled <b><i>train2017_foot_v1.zip</i></b> and <b><i>val2017_foot_v1.zip</i></b>, as provided on this webpage, are owned by Carnegie Mellon University and are licensed under the <a href="https://creativecommons.org/licenses/by/4.0/legalcode">Creative Commons Attribution 4.0 License</a>. All rights not specifically granted are reserved. For any inquiries, please contact Scott McEvoy at <a href = "mailto: [email protected]">[email protected]</a> or <a href = "mailto: [email protected]">[email protected]</a>.</p>
</div> <!-- /container -->
<script src="js/jquery.min.js"></script>
<script src="js/main.js"></script>
<script src="js/videopreview.js"></script>
</body>
</html>