-
Notifications
You must be signed in to change notification settings - Fork 72
/
index_old.html
188 lines (165 loc) · 11.6 KB
/
index_old.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
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN""http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<title>Normalized Object Coordinate Space for Category-Level 6D Object Pose and Size Estimation, CVPR 2019
(Oral)</title>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<!-- Meta tags for Zotero grab citation -->
<meta name="citation_title"
content="Normalized Object Coordinate Space for Category-Level 6D Object Pose and Size Estimation">
<meta name="citation_author" content="Wang, He">
<meta name="citation_author" content="Sridhar, Srinath">
<meta name="citation_author" content="Huang, Jingwei">
<meta name="citation_author" content="Valentin, Julien">
<meta name="citation_author" content="Song, Shuran">
<meta name="citation_author" content="Guibas, Leonidas">
<meta name="citation_publication_date" content="2019">
<meta name="citation_conference_title" content="CVPR 2019">
<!-- <meta name="citation_volume" content="271"> -->
<!-- <meta name="citation_issue" content="20"> -->
<!-- <meta name="citation_firstpage" content="11761"> -->
<!-- <meta name="citation_lastpage" content="11766"> -->
<meta name="citation_pdf_url" content="http://gvv.mpi-inf.mpg.de/projects/VNect/content/VNect_SIGGRAPH2017.pdf">
<!-- Meta tags for search engines to crawl -->
<meta name="robots" content="index,follow">
<meta name="description"
content="The goal of this paper is to estimate the 6D pose and dimensions of unseen object instances in an RGB-D image. Contrary to “instance-level” 6D pose estimation tasks, our problem assumes that no exact object CAD models are available during either training or testing time. To handle different and unseen object instances in a given category, we introduce Normalized Object Coordinate Space (NOCS)—a shared canonical representation for all possible object instances within a category. Our region-based neural network is then trained to directly infer the correspondence from observed pixels to this shared object representation (NOCS) along with other object information such as class label and instance mask. These predictions can be combined with the depth map to jointly estimate the metric 6D pose and dimensions of multiple objects in a cluttered scene. To train our network, we present a new context-aware technique to generate large amounts of fully annotated mixed reality data. To further improve our model and evaluate its performance on real data, we also provide a fully annotated real-world dataset with large environment and instance variation. Extensive experiments demonstrate that the proposed method is able to robustly estimate the pose and size of unseen object instances in real environments while also achieving state-of-the-art performance on standard 6D pose estimation benchmarks.">
<meta name="keywords" content="6D pose estimationl category-level; object detection; instance mask; RGB-D">
<link rel="author" href="http://http://srinathsridhar.com"/>
<!-- Fonts and stuff -->
<!--<link href='http://fonts.googleapis.com/css?family=Open+Sans:400italic,700italic,800italic,400,700,800' rel='stylesheet' type='text/css'>-->
<link href='https://fonts.googleapis.com/css?family=Source+Sans+Pro:400italic,700italic,800italic,400,700,800'
rel='stylesheet' type='text/css'/>
<link rel="stylesheet" type="text/css" href="css/project.css" media="screen"/>
<link rel="stylesheet" type="text/css" media="screen" href="css/iconize.css"/>
<!--<script src="js/google-code-prettify/prettify.js"></script>-->
</head>
<body>
<div id="content">
<div class="section logos">
<a href="http://cvpr2019.thecvf.com/" target="_blank"><img width="28%" height="28%" src="images/cvpr.png" align="left"></a>
<a href="http://ai.stanford.edu/" target="_blank"><img width="4.5%" height="4.5%" src="images/stanford.png" ></a>
<a href="https://ai.google/" target="_blank"><img width="6%" height="6%" src="images/google.png"></a>
<a href="http://www.cs.princeton.edu/" target="_blank"><img width="5%" height="5%"
src="images/princeton.png"></a>
<a href="https://research.fb.com/" target="_blank"><img width="10%" height="10%" src="images/fb.png"></a>
</div>
<div class="section head">
<h1>Normalized Object Coordinate Space for Category-Level 6D Object Pose and Size Estimation</h1>
<div class="authors">
<a href="https://geometry.stanford.edu/member/hewang/index.html" target="_blank">He Wang</a><sup>1</sup>  
<a href="http://srinathsridhar.com/" target="_blank">Srinath Sridhar</a><sup>1</sup>  
<a href="http://stanford.edu/~jingweih/" target="_blank">Jingwei Huang</a><sup>1</sup>  
<a href="https://scholar.google.co.uk/citations?user=pZPD0hMAAAAJ&hl=en" target="_blank">Julien Valentin</a><sup>2</sup>  
<a href="https://shurans.github.io/publication.html" target="_blank">Shuran Song</a><sup>3</sup>  
<a href="https://geometry.stanford.edu/member/guibas/" target="_blank">Leonidas J. Guibas</a><sup>1,4</sup>
</div>
<div class="affiliations">
<sup>1</sup><a href="http://ai.stanford.edu/" target="_blank">Stanford University</a>  
<sup>2</sup><a href="https://ai.google/" target="_blank">Google Inc.</a>  
<sup>3</sup><a href="http://www.cs.princeton.edu/" target="_blank">Princeton University</a>  
<sup>4</sup><a href="https://research.fb.com/" target="_blank">Facebook AI Research</a>  <br/>
</div>
<div class="venue"><a href="http://cvpr2019.thecvf.com/" target="_blank">CVPR 2019</a> [<b>Oral Presentation</b>]
</div>
</div>
<div id="content-inner">
<div class="section abstract">
<div class="section teaser">
<iframe width="640" height="360"
src="https://www.youtube-nocookie.com/embed/UsfoLcInlhM?showinfo=0&controls=1&modestbranding=0&autohide=1&theme=light"
frameborder="0" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>
</div>
<div class="section abstract">
<h2>Abstract</h2>
<p>
The goal of this paper is to estimate the 6D pose and dimensions of unseen object instances in an
RGB-D image. Contrary to “instance-level” 6D pose estimation tasks, our problem assumes that no
exact object CAD models are available during either training or testing time. To handle different
and unseen object instances in a given category, we introduce Normalized Object Coordinate Space
(NOCS)—a shared canonical representation for all possible object instances within a category. Our
region-based neural network is then trained to directly infer the correspondence from observed
pixels to this shared object representation (NOCS) along with other object information such as class
label and instance mask. These predictions can be combined with the depth map to jointly estimate
the metric 6D pose and dimensions of multiple objects in a cluttered scene. To train our network, we
present a new context-aware technique to generate large amounts of fully annotated mixed reality
data. To further improve our model and evaluate its performance on real data, we also provide a
fully annotated real-world dataset with large environment and instance variation. Extensive
experiments demonstrate that the proposed method is able to robustly estimate the pose and size of
unseen object instances in real environments while also achieving state-of-the-art performance on
standard 6D pose estimation benchmarks.
</p>
</div>
<div class="section downloads">
<h2>Downloads</h2>
<center>
<ul>
<li class="grid">
<div class="griditem">
<a href="pub/NOCS_CVPR2019.pdf" target="_blank" class="imageLink"><img
src="images/NOCS_CVPR2019_thumb.png"></a><br/>
Paper<br/><a href="content/NOCS_CVPR2019.pdf" target="_blank">PDF</a>
</div>
</li>
<li class="grid">
<div class="griditem">
<a href="pub/NOCS_CVPR2019_Supp.pdf" target="_blank" class="imageLink"><img
src="images/NOCS_CVPR2019_supp_thumb.png"></a><br/>
Supplementary Document<br/><a href="content/NOCS_CVPR2019_Supp.pdf" target="_blank">PDF</a>
</div>
</li>
<li class="grid">
<div class="griditem">
<a href="https://github.com/hughw19/NOCS_CVPR2019" target="_blank"
class="imageLink">
<img src="images/code.png">
</a><br/>
Code & Datasets<br/>
<a
href="https://github.com/hughw19/NOCS_CVPR2019" target="_blank">on Github</a>
</div>
</li>
</ul>
</center>
</div>
<br/>
<div class="section list">
<h2>Citation</h2>
<p><a href="pub/NOCS_CVPR2019.bib" target="_blank">BibTeX, 1 KB</a></p>
<div class="section bibtex">
<center>
<pre>
@InProceedings{Wang_2019_CVPR,
author = {Wang, He and Sridhar, Srinath and Huang, Jingwei and Valentin, Julien and Song, Shuran and Guibas, Leonidas J.},
title = {Normalized Object Coordinate Space for Category-Level 6D Object Pose and Size Estimation},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2019}
}
</pre>
</center>
</div>
</div>
<div class="section acknowledgments">
<h2>Acknowledgments</h2>
<p>
This research was supported by a grant from Toyota-Stanford Center for AI Research, NSF grant
IIS-1763268, a gift from Google, and a Vannevar Bush Faculty Fellowship. We thank Xin Wang, Shengjun
Qin, Anastasia Dubrovina, Davis Rempe, Li Yi, and Vignesh Ganapathi-Subramanian.
</p>
</div>
<div class="section contact">
<h2>Contact</h2>
Srinath Sridhar<br/><a href="mailto:[email protected]">[email protected]</a>
</div>
<div class="section">
<hr class="smooth">
This page is <a href="http://www.zotero.org" target="_blank">Zotero</a> translator friendly.
<!--Page last updated-->
<!--#config timefmt="%d-%b-%Y" --><!--#echo var="LAST_MODIFIED" -->
</div>
</div>
</div>
</div>
</body>
</html>