-
Notifications
You must be signed in to change notification settings - Fork 2
Expand file tree
/
Copy pathPurNet.html
More file actions
128 lines (127 loc) · 5.81 KB
/
PurNet.html
File metadata and controls
128 lines (127 loc) · 5.81 KB
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
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<title>PurNet</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">
Salient Object Detection with Purificatory
</h1>
<h1 class="title">
Mechanism and Structural Similarity Loss
</h1>
<div class="author">Jia Li</div>
<div class="author">Jinming Su</div>
<div class="author">Changqun Xia</div>
<div class="author">Mingcan Ma</div>
<div class="author">Yonghong Tian</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>
</div>
<div class="con">
<p style="font-size: 24px; margin-top:5px; margin-bottom: 15px;">
IEEE Transactions on Image Processing 2021
</p>
</div>
<div class="columns">
<div class="column"></div>
<div class="column"></div>
<div class="column">
<a href="https://arxiv.org/abs/1912.08393" target="_blank">
<p class="link">Paper</p>
</a>
</div>
<div class="column">
<a href="https://github.com/iCVTEAM/PurNet/" 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/PurNet/head.png" class="centerImage">
</div>
</div>
<div class="head_cap">
<p style="color:gray;">
The framework of PurNet
</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;">
Image-based salient object detection has made great
progress over the past decades, especially after the revival of
deep neural networks. By the aid of attention mechanisms to
weight the image features adaptively, recent advanced deep
learning-based models encourage the predicted results to
approximate the ground-truth masks with as large predictable
areas as possible, thus achieving the state-of-the-art performance.
However, these methods do not pay enough attention to small
areas prone to misprediction. In this way, it is still tough to
accurately locate salient objects due to the existence of regions
with indistinguishable foreground and background and regions
with complex or fine structures. To address these problems, we
propose a novel convolutional neural network with purificatory
mechanism and structural similarity loss. Specifically, in order
to better locate preliminary salient objects, we first introduce
the promotion attention, which is based on spatial and channel
attention mechanisms to promote attention to salient regions.
Subsequently, for the purpose of restoring the indistinguishable
regions that can be regarded as error-prone regions of one model,
we propose the rectification attention, which is learned from the
areas of wrong prediction and guide the network to focus on
error-prone regions thus rectifying errors. Through these two
attentions, we use the Purificatory Mechanism to impose strict
weights with different regions of the whole salient objects and
purify results from hard-to-distinguish regions, thus accurately
predicting the locations and details of salient objects. In addition
to paying different attention to these hard-to-distinguish regions,
we also consider the structural constraints on complex regions
and propose the Structural Similarity Loss. The proposed loss
models the region-level pair-wise relationship between regions
to assist these regions to calibrate their own saliency values. In
experiments, the proposed purificatory mechanism and structural
similarity loss can both effectively improve the performance, and
the proposed approach outperforms 19 state-of-the-art methods
on six datasets with a notable margin. Also, the proposed method
is efficient and runs at over 27FPS on a single NVIDIA 1080Ti GPU.
</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>
Qualitative comparisons
</h1>
</div>
<div class="container" style="max-width:800px">
<div style="text-align: center;">
<img src="assets/PurNet/comp.png" class="centerImage">
</div>
</div>
</div>
</section>
</body>
</html>