<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p><br>
</p>
<p>Hi everyone - <br>
</p>
<p>Please come to our Colloquium today with our speaker Ming-Hsuan
Wang:</p>
<p>
<blockquote type="cite">
<p class="MsoNormal" style="text-align:justify;margin:0in 0in
0.0001pt;font-size:12pt;font-family:Calibri,sans-serif"><i><span
style="font-family:Arial,sans-serif">Host: Tobias Höllerer
& William Wang</span></i></p>
<p class="MsoNormal" style="text-align:justify;margin:0in 0in
0.0001pt;font-size:12pt;font-family:Calibri,sans-serif"><b><span
style="font-family:Arial,sans-serif"> </span></b></p>
<p class="MsoNormal" style="text-align:justify;margin:0in 0in
0.0001pt;font-size:12pt;font-family:Calibri,sans-serif"><b><span
style="font-family:Arial,sans-serif">Title: Learning to
Track and Segment Objects in Videos</span></b><b><span
style="font-family:Arial,sans-serif;color:black"> </span></b></p>
<p class="MsoNormal" style="text-align:justify;margin:0in 0in
0.0001pt;font-size:12pt;font-family:Calibri,sans-serif"><span
style="font-family:Arial,sans-serif"> </span></p>
<p class="MsoNormal" style="text-align:justify;margin:0in 0in
0.0001pt;font-size:12pt;font-family:Calibri,sans-serif"><b><span
style="font-family:Arial,sans-serif">Abstract: </span></b></p>
<span style="font-family:Arial,sans-serif">Visual tracking and
object segmentation are two important tasks for understanding
the motion contents in videos that find numerous applications.
The typical setup for visual tracking is to initialize the
target of interest with a bounding box in the first frame and
the goal is determine locations of the target throughout the
frames. While significant progress has been made in the past
decades, object tracking remains a difficult problem due to
large appearance change, shape deformation, light variation,
and occlusions. In the first part, I will present our recent
findings on tracking single and multiple objects in videos. I
will also discuss numerous applications based on visual
tracking. On the other hand, a closely related problem to how
to segment objects in videos. </span><span
style="font-family:Arial,sans-serif;color:black">Online video
object segmentation is a difficult task as it entails to
process the image sequence timely and accurately. To segment a
target object through the video, numerous methods based on
deep learning have been developed by heavily fine-tuning on
the object mask in the first frame, which is time-consuming
for online applications. In the second part, I will present a
fast and accurate video object segmentation algorithm with
applications. </span><span
style="font-family:Arial,sans-serif">When time allows, I will
briefly discuss our work on low-level and high-level vision
tasks including frame interpolation, image synthesis, and
style transfer. </span></blockquote>
I will bring Ming-Hsuan by the HFH lab around 5pm for a quick tour
and chat - would be great if you could briefly show your work! <br>
</p>
<p>Also, please all go to the student lunch tomorrow (Thu) at noon:
<br>
</p>
<p><a class="moz-txt-link-freetext" href="https://docs.google.com/spreadsheets/d/1mptIBhL8hvEYK8OeFBrBGbahSgym1pvQW-Y8alqgjZI/edit#gid=1873430703">https://docs.google.com/spreadsheets/d/1mptIBhL8hvEYK8OeFBrBGbahSgym1pvQW-Y8alqgjZI/edit#gid=1873430703</a><br>
</p>
<p>Cheers,</p>
<p> Tobias</p>
<p><br>
</p>
<pre class="moz-signature" cols="72">--
Tobias Hollerer
Professor, Department of Computer Science
University of California, Santa Barbara, CA 93106-5110
<a class="moz-txt-link-abbreviated" href="mailto:holl@cs.ucsb.edu">holl@cs.ucsb.edu</a>, Office: (805)284-9395, Fax: (805)893-8553 </pre>
</body>
</html>