-
Notifications
You must be signed in to change notification settings - Fork 0
/
index.html
102 lines (98 loc) · 4.95 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
<!DOCTYPE html>
<html lang="en-US">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Generating depth images of preterm infants with a given pose using GANs</title>
<meta property="og:title" content="Generating depth images of preterm infants with a given pose using GANs">
<meta property="og:url" content=""> <!-- content="infantmotion.github.io" -->
<meta property="og:site_name" content=""> <!-- content="infantmotion.github.io" -->
<link rel="icon" type="image/x-icon" href="./resources/favicon.png">
<link rel="stylesheet" href="./resources/style.css">
</head>
<body>
<h2>Generating depth images of preterm infants with a given pose using GANs</h2>
<p>
<a href="https://it.linkedin.com/in/giuseppe-pio-cannata-b64633158">Giuseppe Pio Cannata<sup>a</sup></a>,
<a href="https://it.linkedin.com/in/lucia-migliorelli-a4571577">Lucia Migliorelli<sup>a</sup></a>,
<a href="https://vrai.dii.univpm.it/adriano.mancini">Adriano Mancini<sup>a</sup></a>,
<a href="https://vrai.dii.univpm.it/emanuele.frontoni">Emanuele Frontoni<sup>b</sup></a>,
<a href="https://www.santannapisa.it/en/sara-moccia">Sara Moccia<sup>c</sup></a>
</p>
<p>
<sup>a</sup> Università Politecnica delle Marche,
<sup>b</sup> Università degli Studi di Macerata,
<sup>c</sup> Scuola Superiore Sant’Anna
</p>
<br>
<h3>Abstract</h3>
<div align="justify">
<p>The use of deep learning (DL) for preterm infant's movement monitoring has the potential to support clinicians
in early recognizing motor and behavioural disorders. The development of DL algorithms is, however, hampered by
the lack of publicly available annotated datasets. In this paper, we present a Generative Adversarial Network
(GAN)-based framework to generate images of preterm infants' in a given pose to partially mitigate this issue.
The framework consists of a bibranch encoder and a conditional GAN, to generate a rough image and a refined
version of it, respectively. Evaluation was performed on the Moving INfants In RGB-D (MINI-RGBD) dataset, which
consists of 12000 depth images acquired in top-view from 12 infants. A low Fréchet inception distance (142.9) and
an inception score (2.8) close to that of real-image distribution (2.6) are obtained, showing the potentiality
of the proposed framework in generating realistic images of preterm infants' in a given pose.
</p>
</div>
<a href="./resources/available_soon.html">
<button type="button" class="btn btn-primary">Download the paper</button>
</a>
<br><br>
<div align="left">
<h2>Proposed GAN-based framework</h2>
</div>
<p>
<img src="./resources/workflow.png" alt="alt text">
</p>
<br>
<p align="justify">
Workflow of the proposed generative adversarial network framework to generate depth
images of preterm infants with a given poses. It's consists of a double-branch convolutional autoencoder
G<sub>1</sub> and a conditional Deep Generativive Adversarial Netowrk (cDCGAN). The framework is fed with:
(i) a depth input image of a preterm infant (I<sub>C</sub>), which acts as condition image,
and (ii) a target pose (P<sub>T</sub>). This is a stack of N images,
where each image is a keypoint binary image built from a target image (I<sub>T</sub>) of a different infant.
<b>The framework was developed using Tensorflow as a backend.</b>
</p>
<a href="https://github.com/vrai-group/guided-infant-generation">
<button type="button" class="btn btn-primary">Go to github</button>
</a>
<br><br>
<div align="left">
<h2 id="results">Results</h2>
</div>
<p align="middle">
<video width="910" height="550" controls="" preload="" muted>
<source src="./resources/video.mp4" type="video/mp4">
</video>
</p>
<br>
<!--
<div align="left">
<h2 id="cite">Cite</h2>
<div>
<div class="language-plaintext highlighter-rouge">
<div class="highlight">
<pre class="highlight">
<code>
</code>
</pre>
</div>
</div>
-->
<p> </p>
<p> </p>
<h3 id="----------">
<a href="https://vrai.dii.univpm.it/"><img src="./resources/vrai.png" width="20%" height="20%"></a>
<a href="https://www.unimc.it/it"><img src="./resources/unimc.png" width="20%" height="20%"></a>
<a href="https://www.santannapisa.it/it"><img src="./resources/scuola_santanna.png" width="25%" height="25%"></a>
</h3>
</body>
</html>