-
Notifications
You must be signed in to change notification settings - Fork 0
/
index.html
138 lines (117 loc) · 8.13 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
131
132
133
134
135
136
137
138
<!-- saved from url=(0042)https://x-ytong.github.io/project/GID.html -->
<html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<title>GID Dataset</title>
<style id="system" type="text/css">
h1,h2,h3,h4,h5,h6,p,blockquote { margin: 0; padding: 0;}
body {font-family: "Helvetica Neue", Helvetica, "Hiragino Sans GB", Arial, sans-serif;
font-size: 15px; line-height: 20px; color: #737373; margin: 10px 13px 10px 13px;}
tr,td {font-size: 15px; color: #737373; text-align: justify; text-justify: inter-character;}
ul {text-align: justify; text-justify: inter-character;}
a {color: #ff7f50; text-decoration: none;}
a:hover {color: #c39797; text-decoration: none;}
a img {border: none;}
p {margin-bottom: 9px; text-align: justify; text-justify: inter-character;}
h1,h2,h3,h4,h5,h6 {color: #20b2aa; line-height: 36px;}
h1 {margin-top: 50px; margin-bottom: 18px; font-size: 26px;}
h2 {margin-top: 24px; font-size: 20px;}
h3 {font-size: 18px;}
h4 {font-size: 16px;}
h5 {font-size: 14px;}
h6 {font-size: 13px;}
hr {margin: 0 0 19px; border: 0; border-bottom: 1px solid #ccc;}
blockquote {padding: 13px 13px 21px 15px; margin-bottom: 18px; font-family:georgia,serif; font-style: italic;}
blockquote:before {content:"C"; font-size:40px; margin-left:-10px; font-family:georgia,serif; color:#eee;}
blockquote p {font-size: 14px; font-weight: 300; line-height: 18px; margin-bottom: 0; font-style: italic;}
code, pre {font-family: Monaco, Andale Mono, Courier New, monospace;}
code {background-color: #fee9cc; color: rgba(0, 0, 0, 0.75); padding: 1px 3px; font-size: 12px;
-webkit-border-radius: 3px; -moz-border-radius: 3px; border-radius: 3px;}
pre {display: block; padding: 14px; margin: 0 0 18px; line-height: 16px; font-size: 11px;
border: 1px solid #d9d9d9; white-space: pre-wrap; word-wrap: break-word;}
pre code {background-color: #fff; color:#737373; font-size: 11px; padding: 0;}
@media screen and (min-width: 768px) {body {width: 748px; margin:10px auto;}}
</style><style id="custom" type="text/css"></style></head>
<body marginheight="0">
<div align="center"><h1>GID-15: A large scale semantic segmentation dataset for remote sensing images</h1></div>
<div style="border:2px solid #FFFFFF "></div>
<!--<div align="center">Xin-Yi Tong, Gui-Song Xia, Qikai Lu, Huanfeng Shen, Shengyang Li, Shucheng You, Liangpei Zhang</div>-->
<h2>Abstract</h2>
<p>In recent years, large amount of high spatial-resolution remote sensing (HRRS) images are available for land-cover mapping.
However, due to the complex information brought by the increased spatial resolution and the data disturbances caused by
different conditions of image acquisition, it is often difficult to find an efficient method for achieving accurate land-cover
classification with high-resolution and heterogeneous remote sensing images.
We create a large scale semantic segmentation dataset for remote sensing images containing <b>150 Gaofen-2 satellite images</b>, 100 images, 10 images and 40 images for training, validating and testing respectively.
<!-- Experiments on multi-source HRRS images, including Gaofen-2, Gaofen-1, Jilin-1, Ziyuan-3, Sentinel-2A, and Google Earth platform data, show encouraging results-->
<!--and demonstrate the applicability of the proposed scheme to land-cover classification with multi-source HRRS images.-->
</p>
<h2>Dataset</h2>
<p>We construct a new large-scale land-cover dataset with Gaofen-2 (GF-2) satellite images. This new dataset, which is named as
Gaofen Image Dataset with 15 categories (GID-15), has superiorities over the existing land-cover dataset because of its large coverage, wide distribution,
and high spatial resolution.
The large-scale remote sensing semantic segmentation set contains 150 pixel-level annotated GF-2 images, which is labeled in 15 categories.
Some of the images are from the paper: <b>Land-Cover Classification with High-Resolution Remote Sensing Images Using Transferable Deep Models</b>
</p>
<h2> Data Compose</h2>
<p>
We will not release the annotation for testing set, so we can get fair comparative results through online benchmark.
In addition, we provide two types of ground truth, '.png' and '.tiff' format respectively.
the '.png' format ground truth is grey label, while '.tiff' format is 'RGB' label, the color palette pls refer to the readme.txt in dataset file.
<!-- the color palette for convert the two types and CLASSES table are as follows:<br>-->
<!--<br>-->
<!--CLASSES table : (1:'industrial_land',2:'urban_residential',3:'rural_residential',4:'traffic_land',5:'paddy_field',<br>-->
<!-- 6:'irrigated_land',7:'dry_cropland',8:'garden_plot',9:'arbor_woodland',10:'shrub_land',11:'natural_grassland',<br>-->
<!-- 12:'artificial_grassland',13:'river',14:'lake',15:'pond')<br>-->
<!-- <br>-->
<!-- In '.png' format GT, the value of each pixel represents a category as we mentioned in CLASSES table, e.g: the pixel value is 1,so the category of this pixel is 'industrial_land'.-->
<!-- It should be mentioned that 0 stands for background, which is not included in 15 categories.<br>-->
<!--<br>-->
<!--PALETTE:<br>-->
<!-- 1:[200, 0, 0],<br>-->
<!-- 2:[250, 0, 150],<br>-->
<!-- 3:[200, 150, 150],<br>-->
<!-- 4:[250, 150, 150],<br>-->
<!-- 5:[0, 200, 0],<br>-->
<!-- 6:[150, 250, 0],<br>-->
<!-- 7:[150, 200, 150],<br>-->
<!-- 8:[200, 0, 200],<br>-->
<!-- 9:[150, 0, 250],<br>-->
<!-- 10:[150, 150, 250],<br>-->
<!-- 11:[250, 200, 0],<br>-->
<!-- 12:[200, 200, 0],<br>-->
<!-- 13:[0, 0, 200],<br>-->
<!-- 14:[0, 150, 200],<br>-->
<!-- 15:[0, 200, 250]<br>-->
</p>
<!--<h3>- Large-scale classification set</h3>-->
<!-- <div align="center"><img src="./GID Dataset_files/GIDlarge.png" width="70%"></div>-->
<h3>- large-scale remote sensing semantic segmentation set</h3>
<div align="center"><img src="./GID Dataset_files/GIDfine.png" width="70%"></div>
<!--<h3>- Multi-source validation images</h3>-->
<!-- <div align="center"><img src="./GID Dataset_files/GIDmulti.png" width="70%"></div>-->
<h3>- Download</h3>
<p>GID-15 can be download from google drive:</p>
<p></p><li>Link: <a href="https://drive.google.com/file/d/1zbkCEXPEKEV6gq19OKmIbaT8bXXfWW6u/view?usp=sharing"><b>google drive</b></a></li><p></p>
<!--<p></p><li>Link: <a href="https://whueducn-my.sharepoint.com/:f:/g/personal/xinyi_tong_whu_edu_cn/EmT3E8xDYaBHklxweG0Je68BGPRWaitir0CuSJaaCjJfDQ?e=87V6PA/"><b>Onedrive (compressed version)</b></a></li><p></p>-->
<!--<p></p><li>Link: <a href="https://pan.baidu.com/s/1kdMdgXCUWFmlpaKXjFRaaA/"><b>Baidudrive</b></a> (extraction code:5r1z)</li><p></p>-->
<!--<h2>Experiment</h2>-->
<!--<p>We test our algorithm and analyse the experimental results in this section. Two types of land-cover classification issues are -->
<!--examined: 1) transferring deep models to classify HRRS images captured with the same sensor and under different conditions, -->
<!--2) transferring deep models to classify multi-source HRRS images. For performance comparison, several object-based land-cover -->
<!--classification methods are utilized.</p>-->
<!--<h3>- Experiments on Gaofen-2 images</h3>-->
<!-- <div align="center"><img src="./GID Dataset_files/GIDresultmap5.png" width="80%"></div>-->
<!-- <div style="border:10px solid #FFFFFF "></div>-->
<!-- <div align="center"><img src="./GID Dataset_files/GIDresultmap15.png" width="80%"></div>-->
<!--<h3>- Experiments on multi-source images</h3>-->
<!-- <div align="center"><img src="./GID Dataset_files/GIDmultiresultmap15.png" width="80%"></div>-->
<h2>Citation</h2>
<pre>@article{GID2020,
title = {Land-Cover Classification with High-Resolution Remote Sensing Images Using Transferable Deep Models},
author = {Xin-Yi Tong, Gui-Song Xia, Qikai Lu, Huangfeng Shen, Shengyang Li, Shucheng You, Liangpei Zhang},
journal = {Remote Sensing of Environment, doi: 10.1016/j.rse.2019.111322},
year = {2020}
}
</pre>
<h2>Contact</h2>
<p>E-mail : [email protected]</p>
<div style="border:20px solid #FFFFFF "></div>
</body></html>