-
Notifications
You must be signed in to change notification settings - Fork 0
/
tasks.html
340 lines (302 loc) · 17.9 KB
/
tasks.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
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no">
<meta name="description" content="">
<meta name="author" content="">
<title>遥感图像智能解译技术挑战赛</title>
<!-- Bootstrap core CSS -->
<link href="vendor/bootstrap/css/bootstrap.min.css" rel="stylesheet">
<!-- Custom fonts for this template -->
<link href="vendor/fontawesome-free/css/all.min.css" rel="stylesheet" type="text/css">
<link href='https://fonts.googleapis.com/css?family=Lora:400,700,400italic,700italic' rel='stylesheet' type='text/css'>
<link href='https://fonts.googleapis.com/css?family=Open+Sans:300italic,400italic,600italic,700italic,800italic,400,300,600,700,800' rel='stylesheet' type='text/css'>
<!-- Custom styles for this template -->
<!-- <link href="css/myclean-blog.min.css" rel="stylesheet"> -->
<!-- <link href="css/myclean-blog.min.css" rel="stylesheet"> -->
<link href="css/myclean-blog.min.css" rel="stylesheet">
<link href="css/common.css" rel="stylesheet">
</head>
<body>
<!-- Navigation -->
<nav class="navbar navbar-expand-lg navbar-light fixed-top" id="mainNav">
<div class="container mContainer">
<button class="navbar-toggler navbar-toggler-right" type="button" data-toggle="collapse" data-target="#navbarResponsive" aria-controls="navbarResponsive" aria-expanded="false" aria-label="Toggle navigation">
Menu
<i class="fas fa-bars"></i>
</button>
<div class="collapse navbar-collapse" id="navbarResponsive">
<ul class="navbar-nav ml-auto m-navbar">
<li class="nav-item">
<a class="nav-link" href="index.html" style="font-size:14px">主页</a>
</li>
<!-- <li class="nav-item">
<a class="nav-link" href="cfp.html" style="font-size:14px">CFP</a>
</li> -->
<!-- <li class="nav-item">
<a class="nav-link" href="people.html" style="font-size:14px">People</a>
</li> -->
<li class="nav-item">
<a class="nav-link" href="dataset.html" style="font-size:14px">数据集</a>
</li>
<li class="nav-item">
<a class="nav-link" href="tasks.html" style="font-size:14px">竞赛任务</a>
</li>
<!--
<li class="nav-item dropdown">
<a class="nav-link dropdown-toggle" href="#" id="navbarDropdown" style="font-size:14px" role="button" data-toggle="dropdown"
aria-haspopup="true" aria-expanded="false">
Challenge
</a>
<div class="dropdown-menu" aria-labelledby="navbarDropdown">
<a class="dropdown-item" href="registration.html">Registration</a>
<a class="dropdown-item" href="dataset.html">Dataset</a>
<a class="dropdown-item" href="tasks.html">Tasks</a>
<a class="dropdown-item" href="evaluation.html">Evaluation</a>
<a class="dropdown-item" href="results.html">Results</a>
</div>
</li> -->
<li class="nav-item">
<a class="nav-link" href="registration.html" style="font-size:14px"> 报名注册 </a>
</li>
<li class="nav-item">
<a class="nav-link" href="program.html" style="font-size:14px">竞赛要求与流程</a>
</li>
<li class="nav-item">
<a class="nav-link" href="results.html" style="font-size:14px"> 测评结果 </a>
</li>
<li class="nav-item">
<a class="nav-link" href="sponsors.html" style="font-size:14px">赞助公司</a>
</li>
<li class="nav-item">
<a class="nav-link" href="contact.html" style="font-size:14px">联系我们</a>
</li>
</ul>
</div>
</div>
</nav>
<header class="masthead m-height" style="background-image: url('images/SECOND_bg.png');">
<div class="overlay" style="background-color: #212529;opacity: 0.85"></div>
<div class="container">
<div class="row">
<div class="col-lg-8 col-md-10 mx-auto">
<div class="site-heading m-head">
<!-- <h2 style="text-align:center; margin-top:1.7em ;font-weight: bold;">
The 1st Workshop on
</h2> -->
<h1 style="text-align:center; margin-top:1.7em; font-weight: bold ; color:#FF9900">
遥感图像智能解译技术挑战赛
</h1>
<h3 style="text-align:center; font-weight: bold; font-style:normal">
第四届中国模式识别与计算机视觉大会
</h3>
</h3 style="text-align:center; font-weight: bold; font-style: italic">
2021-10-29 ~ 2021-11-1 北京国际会议中心
</h3>
<!-- <span class="subheading" style="text-align:center; font-weight:bold; font-style: italic">
</span> -->
</div>
</div>
</div>
</div>
</header>
<!-- Main Content -->
<div class="container">
<div class="row">
<div class="col-lg-10 col-md-10 mx-auto">
<h2>
竞赛的目的与意义
</h2>
<p>
遥感图像在智慧城市、精准农业、紧急救援、灾害评估、国防和公共安全等领域都有重要应用。遥感图像解译是计算机视觉和遥感对地观测的交叉研究领域，举办遥感图像解译竞赛有助于提高遥感图像解译技术的研究水平，推动计算机视觉学科与遥感图像解译产业发展，促进基于人工智能的遥感图像解译技术在遥感领域的应用,培养遥感图像智能解译人才。
</p>
<h2>
<b>赛题：</b>遥感图像建筑物变化检测
<!--<strong>(Recommended)</strong> -->
</h2>
<!-- </br> -->
</br>
<h3>
赛题描述
</h3>
<p>
遥感图像变化检测竞赛即利用同一地理位置前后两个成像时间的遥感图像作为输入，分析其光谱特征以及各地物空间位置分布，提取该地理位置中的变化像元并得到其变化类型。本项竞赛以包含地面建筑物变化的光学遥感图像对作为处理对象，参赛队伍使用主办方提供的遥感图像数据进行变化检测处理，主办方依据评分标准对检测识别结果进行综合评价。
</p>
<!-- <h3>
Evaluation Server
</h3>
<p>
For evaluation, you must registrate and submit on the
<a href="http://www.icdar2017chinese.site:5080/evaluation1/">Evaluation Server</a>
</p> -->
<h3>
提交格式
</h3>
<p>
参赛者需在测试图像对上的每个像素位置赋予变化与未变化两个类别，提交结果的格式为PNG格式的建筑物变化图，每个像素位置的预测结果由数字表示，
其中255代表发生建筑物变化，0代表未发生建筑物变化。
<a style="text-decoration:underline" href="http://47.108.71.49:9005/login/">点击开始测评</a> .
</p>
<!-- <div class="alert alert-secondary" role="alert" style="font-size:18px;font-style: italic;font-family:'Times New Roman', Times, serif">
未变化： (R:0,G:0,B:0)→(R:0,G:0,B:0)
<br> 水 体→建 筑 物： (R:0,G:0,B:255)→(R:128,G:0,B:0)
<br> ...
<br>树 木→建 筑 物： (R:0,G:255,B:0)→(R:128,G:0,B:0)
</div> -->
<h3>
评价指标
</h3>
<p>
<!-- 比赛初赛成绩主要采用SeK系数指标。首先基于土地覆盖变化类型的真实值与预测值计算混淆矩阵 (未变化类别对应首行首列)，剔除混淆矩阵对角线的第一个元素得到矩阵Q (大小为30 x 30)，再利用Kappa系数计算公式得到K<sup>'</sup>。同时，计算Q中正确提取的变化像素所占比例得到IoU<sup>'</sup>，最后根据公式SeK=e<sup>(IoU<sup>'</sup>-1)</sup>×K<sup>'</sup>得到SeK系数指标。SeK值越高，模型的语义变化检测结果越好，排名越高。比赛决赛成绩将基于算法模型精度、效率、规模等指标加权，对算法模型性能进行综合评估与排名。 -->
本次比赛成绩主要采用Overall Accuracy (OA)与mean Intersection Over Union (mIOU) 作为评价指标。首先基于建筑物变化检测的真实标签与预测结果，计算混淆矩阵<i>Q</i>:
<img src="images/math_Q.png" alt="centered image" height="90"> <br>其中，True Positive (TP), False Positive (FP), True Negative (TN)和False Negative (FN)分别代表真阳性样本数，假阳性样本数，真阴性样本数以及假阴性样本数。随后计算得：
<img src="images/oa.png" alt="centered image" height="70">
<br>
<img src="images/miou.png" alt="centered image" height="70"> <br> OA与mIOU值越高，模型的建筑物变化检测结果越好，排名越高。
</p>
<!-- <h2>
<b>Task2</b> - Detection with horizontal bounding boxes
</h2>
<p>
Detecting object with horizontal bounding boxes is usual in many previous contests for object detection. The aim of this task is to accurately localize the instance in terms of horizontal bounding box with (x, y, w, h) format. In the task, the ground
truths for training and testing are generated by calculating the axis-aligned bounding boxes over original annotated bounding boxes.
</p>
<h3>
Submission Format
</h3>
<p>
You will be asked to submit a zip file <a href="example_Task2.zip">(example_Task2.zip)</a> containing results for all test images to evaluate your results. The results are stored in 16 files,
<strong style="color:blue">"Task2_plane.txt, Task2_storage-tank.txt, ..."</strong>, each file contains all the results for a specific category. The format of the results is:
</p>
<div class="alert alert-secondary" role="alert" style="font-size:18px;font-style: italic;font-family:'Times New Roman', Times, serif">
imgname score xmin ymin xmax ymax <br> imgname score xmin ymin xmax ymax <br> ...
</div>
<h3>
Evaluation Protocol
</h3>
<p>
The evaluation protocol for horizontal bounding boxes follows the PASCAL VOC benchmark, which uses mean Average Precision(
<strong>mAP</strong>) as the primary metric.
</p>
<h2>
<b>Task3</b> - Instance Segmentation
</h2>
<p>
The aim of this task is to segment each individual object instance.
</p>
<h3>
Submission Format
</h3>
<p>
The submission format and evaluation protocol follow the MS COCO.
</p>
<h3>
Evaluation Protocol
</h3>
<p>
The evaluation protocol for iSAID is similar to the MS COCO metric. The only difference is that we increase the maximum number of instances for evaluation from 100 to 1000, since iSAID can contain 1,000 instances per image.
</p>
<h2>
<b>Task4</b> - Semantic Segmentation
</h2>
<p>
The aim of this task is to give the semantic category for each pixel in aerial images.
</p>
<h3>
Submission Format
</h3>
<p>
Participants will be asked to submit a zip file containing results for all test images. The by a specific color. For example: expected format for each image is a prediction image with the same resolution. Each category is represented:
</p>
<div class="alert alert-secondary" role="alert" style="font-size:18px;font-style: italic;font-family:'Times New Roman', Times, serif">
Paddyfield : (R : 0;G : 200;B : 0)<br> Urbanresidential : (R : 250;G : 0;B : 150)<br>
</div>
<h3>
Evaluation Protocol
</h3>
<p>
The evaluation protocol adopts the Kappa coefficient.
</p> -->
<!-- <h2>
<b>Task3</b> - Jointly object detection and orientation estimation for movable instances
</h2>
<p>
This task aims to estimate the orientation for movable instances(vehicles, planes, and ships), which is important when applied
to tracking. To make it clear, in this task, each instance's location and orientation is represented
by (x, y, w, h, θ) transferred from {(x
<sub>i</sub>, y
<sub>i</sub>), i = 1,2,3,4}.
</p> -->
<!-- <h3>
Evaluation Server
</h3>
<p>
For evaluation, you must registrate and submit on the
<a href="http://www.icdar2017chinese.site:5080/evaluation1/">Evaluation Server</a>
</p> -->
<!-- <h3>
Submission Format
</h3>
<p>
You will be asked to submit a zip file containing results for all test images to evaluate your results.
The format of the results is:
</p>
<div class="alert alert-secondary" role="alert" style="font-size:18px;font-style: italic;font-family:'Times New Roman', Times, serif">
x y w h Θ category score
<br> x y w h Θ category score
<br> ...
</div>
<h3>
Evaluation Protocol
</h3>
<p>
Note that the evaluation protocol of this task is slightly different from that of Task 2. In Task 2, if the IoU between the
predicted box and ground truth is more than a certain threshold, it is assigned to be true positive(TP).
While in Task3, in addition to requiring IoU to be more than a threshold, the difference in angle is
required to be less than a certain threshold. As a result, the mAP for moveable classes in Task2 is upbound
of Task3. There is a similar metric in PASCAL3D+ which was called Average Viewpoint Precision(
<strong>AVP</strong>).
</p> -->
</div>
</div>
</div>
<hr>
<!-- Footer -->
<footer>
<div class="container">
<div class="row">
<div class="col-lg-8 col-md-10 mx-auto">
<ul class="list-inline text-center">
<li class="list-inline-item">
<a href=" https://twitter.com/intent/tweet?text=https%3A//captain-whu.github.io/PRCV2021">
<span class="fa-stack fa-lg">
<i class="fas fa-circle fa-stack-2x"></i>
<i class="fab fa-twitter fa-stack-1x fa-inverse"></i>
</span>
</a>
</li>
<li class="list-inline-item">
<!-- <a href="https://www.facebook.com/sharer/sharer.php?u=https%3A//captain-whu.github.io/DOAI2019"> -->
<a href="https://facebook.com/sharer.php?display=page&u=https%3A//captain-whu.github.io/PRCV2021">
<span class="fa-stack fa-lg">
<i class="fas fa-circle fa-stack-2x"></i>
<i class="fab fa-facebook-f fa-stack-1x fa-inverse"></i>
</span>
</a>
</li>
</ul>
<p class="copyright text-muted">Copyright ©
<a href="http://www.captain-whu.com">CAPTAIN</a>
</p>
</div>
</div>
</div>
</footer>
<!-- Bootstrap core JavaScript -->
<script src="vendor/jquery/jquery.min.js"></script>
<script src="vendor/bootstrap/js/bootstrap.bundle.min.js"></script>
<!-- Custom scripts for this template -->
<script src="js/clean-blog.min.js"></script>
</body>
</html>