forked from happynear/MTCNN_face_detection_alignment
-
Notifications
You must be signed in to change notification settings - Fork 0
/
index.html
130 lines (108 loc) · 6.88 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
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<!-- saved from url=(0059)http://handtracker.mpi-inf.mpg.de/projects/FastHandTracker/ -->
<html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<title>Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Networks</title>
<!-- Meta tags for search engines to crawl -->
<meta name="robots" content="index,follow">
<meta name="description" content="Face detection and alignment in unconstrained environment are challenging due to various poses, illuminations and occlusions. Recent studies show that deep learning approaches can achieve impressive performance on these two tasks. In this paper, we propose a deep cascaded multi-task framework which exploits the inherent correlation between detection and alignment to boost up their performance. In particular, our framework leverages a cascaded architecture with three stages of carefully designed deep convolutional networks to predict face and landmark location in a coarse-to-fine manner. In addition, we propose a new online hard sample mining strategy that further improves the performance in practice. Our method achieves superior accuracy over the state-of-the-art techniques on the challenging FDDB and WIDER FACE benchmarks for face detection, and AFLW benchmark for face alignment, while keeps real time performance.">
<meta name="keywords" content="face detection; face alignment; cascaded convolutional neural network">
<link rel="author" href="http://kpzhang93.github.io/">
<!-- Fonts and stuff -->
<link href="./support/css" rel="stylesheet" type="text/css">
<link rel="stylesheet" type="text/css" href="./support/project.css" media="screen">
<link rel="stylesheet" type="text/css" media="screen" href="./support/iconize.css">
<script async="" src="./support/prettify.js"></script>
</head>
<body>
<div id="content">
<div id="content-inner">
<div class="section head">
<h1>Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Networks</h1>
<div class="authors">
<a href="http://kpzhang93.github.io/">Kaipeng Zhang<sup>1</sup></a>
<a href="http://personal.ie.cuhk.edu.hk/~zz013/">Zhanpeng Zhang<sup>2</sup></a>
<a href="http://dblp.uni-trier.de/pers/hd/l/Li:Zhifeng">Zhifeng Li<sup>1</sup></a>
<a href="http://mmlab.siat.ac.cn/yuqiao/">Yu Qiao<sup>1</sup></a>
</div>
<div class="affiliations">
<sup>1</sup><a href="http://www.siat.ac.cn/jgsz/kyxt/jcs/yjdy/dmtjc/zxtd/"> Multimedia Research Center, </a>
<a href="http://english.siat.cas.cn/">Shenzhen Institutes of Advanced Technology, </a>
<a href="http://english.cas.cn/">Chinese Academy of Sciences</a>
</br>
<sup>2</sup> <a href="http://mmlab.ie.cuhk.edu.hk/">Multimedia Laboratory, </a>
<a href="http://www.ie.cuhk.edu.hk/">Department of Information Engineering, </a>
<a href="http://www.cuhk.edu.hk">The Chinese University of Hong Kong</a>
</div>
<div class="venue">IEEE Signal Processing Letters (<a href="http://signalprocessingsociety.org/publications-resources/ieee-signal-processing-letters/ieee-signal-processing-letters" target="_blank">SPL</a>), vol. 23, no. 10, pp. 1499-1503, 2016</div>
</div>
</br>
<center><img src="./support/index.png" border="0" width="95%"></center>
<div class="section abstract" style="clear: both;text-align:justify; text-justify:inter-ideograph;">
<h2>Abstract</h2></br>
<p>
Face detection and alignment in unconstrained environment are challenging due to various poses, illuminations and occlusions. Recent studies show that deep learning approaches can achieve impressive performance on these two tasks. In this paper, we propose a deep cascaded multi-task framework which exploits the inherent correlation between detection and alignment to boost up their performance. In particular, our framework leverages a cascaded architecture with three stages of carefully designed deep convolutional networks to predict face and landmark location in a coarse-to-fine manner. In addition, we propose a new online hard sample mining strategy that further improves the performance in practice. Our method achieves superior accuracy over the state-of-the-art techniques on the challenging FDDB and WIDER FACE benchmarks for face detection, and AFLW benchmark for face alignment, while keeps real time performance.
</p>
</div>
<div class="section downloads">
<h2>Results</h2><center></br>
<center><img src="./paper/examples.png" border="0" width="100%"></center></br>
<center><img src="./paper/result.png" border="0" width="100%"></center>
</div></center>
<div class="section downloads">
<!--
<h2>Demo</h2><center>
<iframe width="560" height="315" src="" frameborder="0" allowfullscreen></iframe>
</div></center>
<div class="section downloads">
!-->
<h2>Downloads</h2></br>
<center>
<ul>
<li class="grid">
<div class="griditem">
<a href="./paper/spl.pdf" target="_blank" class="imageLink"><img src="./paper/paper.png"></a><br>
Paper<br><a href="./paper/spl.pdf">PDF, 5.1 MB</a>
</div>
</li>
<li class="grid">
<div class="griditem">
<a href="./code/codes.zip" target="_blank" class="imageLink"><img src="./code/code.png"></a><br>
Codes<br><a href="./code/codes.zip" target="_blank">ZIP, 12.1 MB</a>
(<a href=" https://github.com/kpzhang93/MTCNN_face_detection_alignment">Github</a>)
</div>
</ul>
</center>
</div>
<br>
<div class="section list">
<h2>Citation</h2>
<div class="section bibtex">
<pre>@ARTICLE{7553523,
author={K. Zhang and Z. Zhang and Z. Li and Y. Qiao},
journal={IEEE Signal Processing Letters},
title={Joint Face Detection and Alignment Using Multitask Cascaded Convolutional Networks},
year={2016},
volume={23},
number={10},
pages={1499-1503},
keywords={Benchmark testing;Computer architecture;Convolution;Detectors;Face;Face detection;Training;Cascaded convolutional neural network (CNN);face alignment;face detection},
doi={10.1109/LSP.2016.2603342},
ISSN={1070-9908},
month={Oct},}
</pre>
</div>
</div>
<div class="section contact">
<h2>License</h2></br>
This code is distributed under <a href="https://kpzhang93.github.io/MTCNN_face_detection_alignment/LICENSE">MIT LICENSE</a></br>
</div>
<div class="section contact">
<h2>Contact</h2></br>
Yu Qiao<br><a href="mailto:yu.qiao@siat.ac.cn">yu.qiao@siat.ac.cn</a></br>
Kaipeng Zhang<br><a href="mailto:kpzhang@cmlab.csie.ntu.edu.tw">kpzhang@cmlab.csie.ntu.edu.tw</a></br>
We look forward your sharing of implementation with better runtime efficiency.
</div>
</div>
</div>
<div align="center"><a href="http://www.amazingcounters.com"><img border="0" src="http://cc.amazingcounters.com/counter.php?i=3207443&c=9622642" alt="AmazingCounters.com"></a></div>
</body></html>