iCVTEAM.github.io/ADSOI.html at master · iCVTEAM/iCVTEAM.github.io · GitHub

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
<!DOCTYPE html>
<html>
  <head>
    <meta charset="utf-8">
    <title>ADSOI</title>
    <link rel="stylesheet" type="text/css" href="assets/scripts/bulma.min.css">
    <link rel="stylesheet" type="text/css" href="assets/scripts/theme.css">
    <link rel="stylesheet" type="text/css" href="https://cdn.bootcdn.net/ajax/libs/font-awesome/4.7.0/css/font-awesome.min.css">
  </head>
  <body>
    <section class="hero is-light" style="">
      <div class="hero-body" style="padding-top: 50px;">
        <div class="container" style="text-align: center;margin-bottom:5px;">
          <h1 class="title">
            Attentive Deep Stitching and Quality Assessment for
          </h1>
          <h1 class="title">
            360° Omnidirectional Images
          </h1>

          <div class="author">Jia Li<sup>1</sup></div>
          <div class="author">Yifan Zhao<sup>1</sup></div>
          <div class="author">Weihua Ye<sup>1</sup></div>
          <div class="author">Kaiwen Yu<sup>1</sup></div>
          <div class="author">Shiming Ge<sup>2</sup></div>
          <div class="group">
            <a href="http://cvteam.net/">CVTEAM</a>
          </div>
          <div class="aff">
            <p><sup>1</sup>State Key Laboratory of Virtual Reality Technology and Systems, SCSE, Beihang University, Beijing, China</p>
            <p><sup>2</sup> Institute of Information Engineering, Chinese Academy of Sciences</p>
          </div>
          <div class="con">
            <p  style="font-size: 24px; margin-top:5px; margin-bottom: 15px;">
            JSTSP 2020
            </p>
          </div>
          <div class="columns">
            <div class="column"></div>
            <div class="column"></div>
            <div class="column">
              <a href="http://cvteam.net/papers/2020-JSTSP-Attentive%20Deep%20Stitching%20and%20Quality%20Assessment%20for%20360%C2%B0%20Omnidirectional%20Images.pdf" target="_blank">
                <p class="link">Paper</p>
              </a>
            </div>
            <div class="column">
              <a href="https://github.com/iCVTEAM/ADSOI/" target="_blank">
                <p class="link">Code</p>
              </a>
            </div>
            <div class="column"></div>
            <div class="column"></div>
          </div>
        </div>
      </div>
    </section>
    <div style="text-align: center;">
      <div class="container" style="max-width:850px">
        <div style="text-align: center;">
          <img src="assets/ADSOI/head.png" class="centerImage">
        </div>
      </div>
      <div class="head_cap">
        <p style="color:gray;">
          Framework of proposed Attentive Deep Stitching (ADS).
        </p>
      </div>
    </div>
    <section class="hero">
      <div class="hero-body">
        <div class="container" style="max-width: 800px" >
          <h1 style="">Abstract</h1>
          <p  style="text-align: justify; font-size: 17px;">
            360° omnidirectional images are very helpful in creating
            immersive multimedia contents, which enables a huge demand
            in their efficient generation and effective assessment. In this paper,
            we leverage an attentive idea to meet this demand by addressing two
            concerns: how to generate a good omnidirectional image in a fast
            and robust way and what is a good omnidirectional image for human.
            To this end, we propose an attentive deep stitching approach
            to facilitate the efficient generation of omnidirectional images,
            which is composed of two modules. The low-resolution deformation
            module aims to learn the deformation rules from dual-fisheye to
            omnidirectional images with joint implicit and explicit attention
            mechanisms, while the high-resolution recurrence module enhances the
            resolution of stitching results with the high-resolution guidance in a
            recurrent manner. In this way, the stitching approach can efficiently
            generate high-resolution omnidirectional images that are highly
            consistent with human immersive experiences. Beyond the efficient
            generation, we further present an attention-driven omnidirectional
            image quality assessment (IQA) method which uses joint evaluation
            with both global and local metrics. Especially, the local metric
            mainly focuses on the stitching region and attention region that
            mostly affect the Mean Opinion Score (MOS), leading to a consistent
            evaluation of human perception. To verify the effectiveness of
            our proposed assessment and stitching approaches, we construct a
            hybrid benchmark evaluation with 7 stitching models and 8 IQA
            metrics. Qualitative and quantitative experiments show our stitching
            approach generate preferable results with the state-of-the-art
            models at a 6× faster speed and the proposed quality assessment
            approach surpasses other methods by a large margin and is highly
            consistent with human subjective evaluations.
          </p>
        </div>
      </div>
    </section>
    <section class="hero is-light" style="background-color:#FFFFFF;">
      <div class="hero-body">
        <div class="container" style="max-width:800px;margin-bottom:20px;">
          <h1>
            Benchmark evaluations.
          </h1>
        </div>
        <div class="container" style="max-width:800px">
          <div style="text-align: center;">
            <img src="assets/ADSOI/comp.png" class="centerImage">
          </div>
        </div>
      </div>
    </section>
  <section class="hero" style="padding-top:0px;">
    <div class="hero-body">
      <div class="container" style="max-width:800px;">
  <div class="card">
  <header class="card-header">
    <p class="card-header-title">
      BibTex Citation
    </p>

    <a class="card-header-icon button-clipboard" style="border:0px; background: inherit;" data-clipboard-target="#bibtex-info" >
      <i class="fa fa-copy" height="20px"></i>
    </a>
  </header>
    <div class="card-content">
<pre style="background-color:inherit;padding: 0px;" id="bibtex-info">@article{li2019attentive,
  title={Attentive Deep Stitching and Quality Assessment for 360° Omnidirectional Images},
  author={Li, Jia and Zhao, Yifan and Ye, Weihua and Yu, Kaiwen and Ge, Shiming},
  journal={IEEE Journal of Selected Topics in Signal Processing},
  volume={14},
  number={1},
  pages={209--221},
  year={2019},
  publisher={IEEE}
}</pre>
    </div>
    </section>
    <script type="text/javascript" src="assets/scripts/clipboard.min.js"></script>
    <script>
      new ClipboardJS('.button-clipboard');
    </script>
  </body>
</html>