-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathtemplate.html
60 lines (59 loc) · 3.78 KB
/
template.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
<!doctype html>
<html>
<head>
<title>Convolutional Pose Machines</title>
<link href="css/main.css" media="screen" rel="stylesheet" type="text/css"/>
<script src="js/jquery/jquery.min.js" type="text/javascript"></script>
<script src="js/main.js" type="text/javascript"></script>
</head>
<body>
<div id="container">
<div id="header">
<div id="header-titles">
Convolutional Pose Machines: <br/>
A Deep Architecture with Intermediate Supervision
</div>
<div id="header-arthors">
Shih-En Wei <a href="http://www.cs.cmu.edu/~vramakri/">Varun Ramakrishna</a> Yaser Sheikh
</div>
<div id="header-institute">
The Robotics Institute Carnegie Mellon University
</div>
</div>
<div id="video">
<h3 id="video-title">Demo Videos (frame by frame detection)</h3>
<p id="video-names">
<a href="javascript:void(0)" data-value="1" id="s1" class="video-item">Roger and Rafa</a> |
<a href="javascript:void(0)" data-value="2" id="s2" class="video-item">Dancing Cop</a> |
<a href="javascript:void(0)" data-value="3" id="s3" class="video-item">Freestyle</a> |
<a href="javascript:void(0)" data-value="4" id="s4" class="video-item">Ronald</a> |
<a href="javascript:void(0)" data-value="5" id="s5" class="video-item">Safe Cycling</a>
</p>
<p id="video-explanation">Model trained on LEEDS Sport Dataset with observer centric annotation</p>
<div id="video-container">
<video class="vinstance" controls="controls" src="video/ConvPoseMachine.mp4" width="860"></video>
<!--video class="vinstance" controls="controls" src="video/cop_video2.m4v" height="400"></video>-->
<!--video class="vinstance" controls="controls" src="video/freestyle.mp4" height="400"></video>
<video class="vinstance" controls="controls" src="video/ronaldinho_new.mov" height="400"></video>
<video class="vinstance" controls="controls" src="video/safe_cycling2.m4v" height="400"></video>-->
</div>
</div>
<div id="body">
<div id="body-abstract">
<h3>Overview</h3>
<p><a href="http://www.cs.cmu.edu/~vramakri/poseMachines.html">Pose Machines</a> provide a powerful modular framework for articulated pose estimation.
The sequential prediction framework allows for the learning of rich implicit spatial models, but relies on manually designed features for representing image and spatial context.</p>
</p>In this work, we incorporate a convolutional network architecture into the pose machine framework allowing the learning of representations for both image and spatial context directly from data.
The multiple modules, or levels and stages, in the proposed system can be trained jointly, given that the overall system can be seen as a deep network.
In the same time, our approach addresses the characteristic difficulty of vanishing gradients
during training by providing a natural learning objective function that enforces intermediate supervision, thereby replenishing back-propagated gradients and conditioning the learning procedure.
We have achieved state-of-the-art performance and outperform competing methods on standard
benchmarks: <a href="http://vision.grasp.upenn.edu/cgi-bin/index.php?n=VideoLearning.FLIC">FLIC</a> and <a href="http://www.comp.leeds.ac.uk/mat4saj/lsp.html">LEEDS Sport Dataset</a>.</p>
</div>
<div id="body-img">
<img src="img/teaser.png" height="455" />
</div>
</div>
</div>
</body>
</html>