-
Notifications
You must be signed in to change notification settings - Fork 18
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
0 parents
commit 0d4c9fb
Showing
12 changed files
with
9,023 additions
and
0 deletions.
There are no files selected for viewing
Large diffs are not rendered by default.
Oops, something went wrong.
Large diffs are not rendered by default.
Oops, something went wrong.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Large diffs are not rendered by default.
Oops, something went wrong.
Large diffs are not rendered by default.
Oops, something went wrong.
Binary file not shown.
Binary file not shown.
Large diffs are not rendered by default.
Oops, something went wrong.
Large diffs are not rendered by default.
Oops, something went wrong.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,240 @@ | ||
<!-- saved from url=(0047)https://www.cs.cmu.edu/~peiyunh/tiny/index.html --> | ||
<html xmlns="http://www.w3.org/1999/xhtml"> | ||
|
||
<head> | ||
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> | ||
<link rel="StyleSheet" href="./cvpr21rigidmask/style.css" type="text/css" media="all"> | ||
|
||
<title>Learning to Segment Rigid Motions from Two Frames</title> | ||
<style type="text/css"> | ||
body { | ||
font-family: Arial; | ||
background-color: #f2f2f2; | ||
font-size: 15px; | ||
} | ||
|
||
.content { | ||
width: 800px; | ||
padding: 25px 25px; | ||
margin: 25px auto; | ||
background-color: #fff; | ||
border-radius: 20px; | ||
} | ||
|
||
.description { | ||
font-family: "Times"; | ||
white-space: pre; | ||
text-align: left; | ||
} | ||
|
||
.content-title { | ||
background-color: inherit; | ||
margin-top: 5px; | ||
padding-top: 5px; | ||
margin-bottom: 0; | ||
padding-bottom: 0; | ||
} | ||
|
||
a, | ||
a:visited { | ||
text-decoration: none; | ||
color: blue; | ||
} | ||
|
||
.anchor { | ||
color: inherit; | ||
} | ||
|
||
#authors { | ||
text-align: center; | ||
} | ||
|
||
#conference { | ||
text-align: center; | ||
font-style: italic; | ||
} | ||
|
||
#authors a { | ||
margin: 0 10px; | ||
} | ||
|
||
h1 { | ||
text-align: center; | ||
font-family: Arial; | ||
font-size: 30px; | ||
} | ||
|
||
h2 { | ||
font-family: Arial; | ||
font-size: 25px; | ||
padding: 0; | ||
margin: 10px; | ||
} | ||
|
||
h3 { | ||
font-family: Arial; | ||
font-size: 20px; | ||
padding: 0; | ||
margin: 10px; | ||
} | ||
|
||
p { | ||
line-height: 130%; | ||
margin: 10px; | ||
} | ||
|
||
li { | ||
margin: 10px 0; | ||
} | ||
|
||
.samples { | ||
float: left; | ||
width: 50%; | ||
text-align: center; | ||
} | ||
|
||
.cond { | ||
float: left; | ||
margin: 0 40px; | ||
} | ||
|
||
.cond-container { | ||
width: 700px; | ||
margin: 0 auto; | ||
text-align: center; | ||
} | ||
</style> | ||
|
||
</head> | ||
|
||
|
||
|
||
<body> | ||
|
||
<div class="content content-title" style="text-align: center"> | ||
<h1>Learning to Segment Rigid Motions from Two Frames</h1> | ||
<big style="color:grey;"> | ||
CVPR 2021 | ||
</big> | ||
<p id="authors"> | ||
<table align="center" style="width:60%; text-align:center; table-layout: fixed"> | ||
<tr> | ||
<th><a href="https://gengshan-y.github.io/">Gengshan Yang<sup>1</sup></a></th> | ||
<th><a href="http://www.cs.cmu.edu/~deva/">Deva Ramanan<sup>1,2</sup></a></th> | ||
</tr> | ||
</table> | ||
<sup>1</sup>Robotics Institute, Carnegie Mellon University<br> | ||
<sup>2</sup>Argo AI | ||
</p> | ||
<p> | ||
</p> | ||
</div> | ||
|
||
|
||
|
||
|
||
<div class="content"> | ||
<figure style="font-style: italic; font-weight: normal; margin: 0px; padding: 0px; border: 0px; text-align: left"> | ||
<video autoplay controls loop muted width="810" height="320"> | ||
<source src="./cvpr21rigidmask/rigidmask-teaser.mp4" type="video/mp4"> | ||
</video> | ||
<br> | ||
<figcaption> We propose a neural architecture powered by geometric reasoning that decomposes two frames into a | ||
rigid background and multiple moving rigid bodies, parameterized by 3D rigid transformations and depth. | ||
</figure> | ||
</div> | ||
|
||
|
||
|
||
|
||
|
||
|
||
<div class="content"> | ||
<h2>Abstract</h2> | ||
<p> | ||
Appearance-based detectors achieve remarkable performance on common scenes, benefiting from high-capacity models | ||
and massive annotated data, but tend to fail for scenarios that lack training data. Geometric motion segmentation | ||
algorithms, however, generalize to novel scenes, but have yet to achieve comparable performance to | ||
appearance-based ones, due to noisy motion estimations and degenerate motion configurations. To combine the best | ||
of both worlds, we propose a modular network, whose architecture is motivated by a geometric analysis of what | ||
independent object motions can be recovered from an ego-motion field. It takes two consecutive frames as input and | ||
predicts segmentation masks for the background and multiple rigidly moving objects, which are then parameterized | ||
by 3D rigid transformations. Our method achieves state-of-the-art performance for rigid motion segmentation on | ||
KITTI and Sintel. The inferred rigid motions lead to a significant improvement for depth and scene flow | ||
estimation. | ||
</p> | ||
<div id="teaser" style="margin: 12px; text-align: left;border-top: 1px solid lightgray;padding-top: 12px;"> | ||
<a | ||
href="https://openaccess.thecvf.com/content/CVPR2021/papers/Yang_Learning_To_Segment_Rigid_Motions_From_Two_Frames_CVPR_2021_paper.pdf"> | ||
<strong>[Paper]</strong> | ||
</a> | ||
<a | ||
href="https://openaccess.thecvf.com/content/CVPR2021/supplemental/Yang_Learning_To_Segment_CVPR_2021_supplemental.pdf"> | ||
<strong>[Supp]</strong> | ||
</a> | ||
<a href="https://docs.google.com/presentation/d/1AfF-zPKAWenPXUnFqTQmJaSgIQ90kfhse39d950za_k/edit?usp=sharing"> | ||
<strong>[Slides]</strong> | ||
</a> | ||
<a href="./cvpr21rigidmask/rigidmask-poster.pdf"> | ||
<strong>[Poster]</strong> | ||
</a> | ||
</div> | ||
</div> | ||
|
||
<div class="content"> | ||
<h2>Bibtex</h2> | ||
<p class="description"> | ||
@inproceedings{yang2021rigidmask, | ||
title={Learning to Segment Rigid Motions from Two Frames}, | ||
author={Yang, Gengshan and Ramanan, Deva}, | ||
booktitle={CVPR}, | ||
year={2021} | ||
} | ||
</p> | ||
</div> | ||
|
||
|
||
<div id="vid" class="content"> | ||
<h2>Video</h2> | ||
<iframe width="560" height="315" src="https://www.youtube.com/embed/lsPpM6AjXqA" title="YouTube video player" | ||
frameborder="0" | ||
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" | ||
allowfullscreen></iframe> | ||
<div id="teaser" style="margin: 12px; text-align: left;border-top: 1px solid lightgray;padding-top: 12px;"> | ||
</div> | ||
</div> | ||
|
||
<div class="content"> | ||
<div style="float: right; width:70px; margin-top: 0px; margin-bottom: 25px"> | ||
<a href="https://github.com/gengshan-y/"> | ||
<img src="./github.png" width="100%" alt="github"> | ||
</a> | ||
</div> | ||
<h2>Code</h2> | ||
<p> | ||
Code is available <a href="https://github.com/gengshan-y/rigidmask">here.</a> | ||
</p> | ||
</div> | ||
|
||
<div class="content"> | ||
<h2>Acknowledgments</h2> | ||
<p>This work was supported by the <a href="https://labs.ri.cmu.edu/argo-ai-center/">CMU Argo AI Center for | ||
Autonomous Vehicle Research</a>. We thank Rui Zhu for proving the code of single-image camera intrinsics | ||
estimation. We thank Jason Zhang, Tarasha Khurana, Jessica Lee and many others for their useful feedback.</p> | ||
</div> | ||
|
||
<table width="100%" align="center" border="0" cellspacing="0" cellpadding="20"> | ||
<tr> | ||
<td> | ||
<p align="right"> | ||
<font size="2"> | ||
<a href="https://www.cs.cmu.edu/~peiyunh/">Webpage design borrowed from Peiyun Hu</a> | ||
</font> | ||
</p> | ||
</td> | ||
</tr> | ||
</table> | ||
|
||
</body> | ||
|
||
</html> |
Binary file not shown.