<div dir="ltr">Team,<div><br></div><div>Please come see Chao give his thesis proposal presentation this Friday at 2pm.</div><div><br></div><div>Cheers</div><div>Artur<br><br><div class="gmail_quote"><div dir="ltr">---------- Forwarded message ---------<br>From: <strong class="gmail_sendername" dir="auto">Suzanne Muth</strong> <span dir="ltr"><<a href="mailto:lyonsmuth@cmu.edu">lyonsmuth@cmu.edu</a>></span><br>Date: Wed, Nov 21, 2018 at 10:07 AM<br>Subject: RI Ph.D. Thesis Proposal: Chao Liu<br>To: <a href="mailto:ri-people@cs.cmu.edu">ri-people@cs.cmu.edu</a> <<a href="mailto:ri-people@cs.cmu.edu">ri-people@cs.cmu.edu</a>><br></div><br><br>
<div dir="ltr">
<div id="m_5654550622091363606divtagdefaultwrapper" style="font-size:12pt;color:#000000;font-family:Calibri,Helvetica,sans-serif" dir="ltr">
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Date: 30 November 2018</span></font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Time: 2:00 p.m.</span></font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Place: NSH 4305</span></font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Type: Ph.D. Thesis Proposal</span></font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Who: Chao Liu </span></font></p>
<p style="color:rgb(33,33,33);font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
<span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Topic: </span><font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="color:rgb(34,34,34);font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Vision
with Small Baselines</span></font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><br>
</font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
<span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Abstract:</span><font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><br>
</font></p>
<p class="m_5654550622091363606xmsonormal" style="margin:0in 0in 0.0001pt;font-size:medium;font-family:"Times New Roman",serif">
<span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">Portable camera sensor systems are becoming more and more popular in computer </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">vision
applications such as autonomous driving, virtual reality, robotics manipulation </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">and
surveillance, due to the decreasing expense and size of RGB camera. </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">Despite the
compactness and portability of the small baseline vision systems, it is </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">well-known
that the uncertainty in range finding using multiple views and the sensor </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">baselines
are inversely related. For small baseline vision systems, this means </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">high depth
uncertainties even for close range objects. On the other hand, besides </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">compactness,
small baseline vision system has its unique advantages such as easier </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">correspondence
and large overlapping regions across views. How to utilize those </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">advantages for
small baseline vision setup while avoiding the limitations as much </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">as possible?
In this thesis proposal, we approach this question in terms of three </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">aspects: scene
complexity, uncertainties in the estimations and baseline distance in </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">the setup.</span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"><u></u><u></u></span></p>
<p class="m_5654550622091363606xmsonormal" style="margin:0in 0in 0.0001pt;font-size:medium;font-family:"Times New Roman",serif">
<span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)"> </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"><u></u><u></u></span></p>
<p class="m_5654550622091363606xmsonormal" style="margin:0in 0in 0.0001pt;font-size:medium;font-family:"Times New Roman",serif">
<span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">We first present a method for matting and depth recovery of 3D thin structures </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">with
self-occlusions using a single-view camera with finite aperture lens. In this </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">work,
we take advantage of the small camera baselines that makes the correspondence </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">easier.
We apply the proposed method to scenes at both macro and micro</span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">scales. For macro-scale,
we evaluate our method on scenes with complex 3D thin </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">structures such as tree branches
and grass. For micro-scale, we apply our method to </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)"><i>in-vivo</i> microscopic images
of micro-vessels with diameters less than 50 <i>um</i>.</span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"><u></u><u></u></span></p>
<p class="m_5654550622091363606xmsonormal" style="margin:0in 0in 0.0001pt;font-size:medium;font-family:"Times New Roman",serif">
<span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)"> </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"><u></u><u></u></span></p>
<p class="m_5654550622091363606xmsonormal" style="margin:0in 0in 0.0001pt;font-size:medium;font-family:"Times New Roman",serif">
<span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">We also utilize the small baselines for circularly placed point light sources (commonly </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">seen
in consumer devices like NESTcam, Amazon Cloudcam). We propose </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">a two-stage near-light
photometric stereo method. In the first stage, we optimize</span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">the vertex positions using
the differential images induced by small changes in light</span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">source position. This procedure
yields a strong initial guess for the second stage that </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">refines the estimations
using the raw captured images.</span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"><u></u><u></u></span></p>
<p class="m_5654550622091363606xmsonormal" style="margin:0in 0in 0.0001pt;font-size:medium;font-family:"Times New Roman",serif">
<span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)"> </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"><u></u><u></u></span></p>
<p class="m_5654550622091363606xmsonormal" style="margin:0in 0in 0.0001pt;font-size:medium;font-family:"Times New Roman",serif">
<span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">To handle the estimation uncertainties inherent in the small baseline setup, we </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">propose
a learning-based method to estimate per-pixel depth and its uncertainty continuously </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">from
a monocular video stream. Compared to prior work, the proposed approach </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">achieves
more accurate and stable results, generalizes better to new datasets, </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">and yields
per-pixel depth probability map that accounts for the estimation uncertainties </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">due
to specular surface, occlusions in the scene and objects with large distance.</span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"><u></u><u></u></span></p>
<p class="m_5654550622091363606xmsonormal" style="margin:0in 0in 0.0001pt;font-size:medium;font-family:"Times New Roman",serif">
<span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)"> </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"><u></u><u></u></span></p>
<p class="m_5654550622091363606xmsonormal" style="margin:0in 0in 0.0001pt;font-size:medium;font-family:"Times New Roman",serif">
<span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">To deal with the subsurface light scattering in the tissue, we propose a projector-camera </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">setup
with small baseline that works in a small scale and a method that </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">combines the approximated
model for subsurface light scattering, in order to see </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">through skins and perform
<i>in-vivo</i> blood flow analysis on human skin.</span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"><u></u><u></u></span></p>
<p class="m_5654550622091363606xmsonormal" style="margin:0in 0in 0.0001pt;font-size:medium;font-family:"Times New Roman",serif">
<span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)"> </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"><u></u><u></u></span></p>
<p class="m_5654550622091363606xmsonormal" style="margin:0in 0in 0.0001pt;font-size:medium;font-family:"Times New Roman",serif">
<span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">We also propose to combine the benefits of small and large baseline vision systems, </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">in
order to handle large region occlusion and depth estimation for fine-grained </span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"></span><span style="font-family:Calibri,sans-serif;color:rgb(33,33,33)">structures
at the same time.</span><span style="font-size:11.5pt;font-family:Tahoma,sans-serif;color:rgb(33,33,33)"><u></u><u></u></span></p>
<p class="m_5654550622091363606x_MsoNormal" style="color:rgb(33,33,33);font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px;margin:0in 0in 0.0001pt">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt"></span></font></p>
<p class="m_5654550622091363606x_MsoNormal" style="color:rgb(33,33,33);font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px;margin:0in 0in 0.0001pt">
<br>
</p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Thesis Committee Members:</span><br>
</font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px;margin-right:0px;margin-left:0px;line-height:normal">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Srinivasa Narasimhan, Co-chair</span></font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px;margin-right:0px;margin-left:0px;line-height:normal">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Artur Dubrawski, Co-chair</span></font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px;margin-right:0px;margin-left:0px;line-height:normal">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Aswin Sankaranarayanan </span></font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
</p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px;margin-right:0px;margin-left:0px;line-height:normal">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">Manmohan Chandraker, University of California, San Diego</span></font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
</p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
</p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
</p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
<br>
</p>
<p class="m_5654550622091363606x_MsoNormal" style="color:rgb(33,33,33);font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px;margin:0in 0in 0.0001pt">
<font style="font-family:arial,helvetica,sans-serif,serif,EmojiFont"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">A copy of thesis document is available at:</span></font></p>
<p style="font-family:wf_segoe-ui_normal,"Segoe UI","Segoe WP",Tahoma,Arial,sans-serif,serif,EmojiFont;font-size:15px">
<a href="https://www.dropbox.com/s/wwxe7mqy7mf947q/small-baseline.pdf?dl=0" rel="noopener noreferrer" style="font-family:arial,helvetica,sans-serif" id="m_5654550622091363606LPlnk385183" target="_blank"><span style="font-family:Calibri,Helvetica,sans-serif;font-size:12pt">https://www.dropbox.com/s/wwxe7mqy7mf947q/small-baseline.pdf?dl=0</span></a></p>
<div id="m_5654550622091363606Signature">
<div name="divtagdefaultwrapper">
<font size="2"></font></div>
</div>
</div>
</div>
</div></div></div>