<td align="center" width="14%" style = "vertical-align: middle">
<img src = "/images/oracle.png" width="50%">
</td>
<td align="center" width="14%" style = "vertical-align: middle">
<img src = "/images/infineon.png" width="60%">
</td>
</tr>
<tr>
<td align="center" style = "vertical-align: middle; background-color: rgba(255, 255, 255, 1)">MS, CS<br>Syracuse University<br>2016 - 2018</td>
<td align="center" style = "vertical-align: middle; background-color: rgba(255, 255, 255, 1)">IT Consultant<br>Oracle<br>2012 - 2016</td>
<td align="center" style = "vertical-align: middle; background-color: rgba(255, 255, 255, 1)">Research Intern<br>Infineon<br>2011 - 2012</td>
<!-- <td align="center" style = "vertical-align: middle; background-color: rgba(255, 255, 255, 1)">MTech<br>R.V. College of Engineering<br>2010 - 2012</td>
<td align="center" style = "vertical-align: middle; background-color: rgba(255, 255, 255, 1)">BE<br>M.S.Ramaiah Institute of Technology<br>2005 - 2009</td> -->
</tr>
</table>
</div>
<!-- Research Interest -->
<table width="100%" align="center" border="0" cellspacing="0" cellpadding="20">
<tr>
<td width="100%" valign="middle">
<heading>Research</heading>
<p align="justify">
My research interests are to develop deep learning algorithms for 3D computer vision problems and create end-to-end solution pipelines. My long-term goal is to build a mobile-based assistant for the visually impaired to help them navigate the real world.
</p>
</td>
</tr>
</table>
<!-- Publications-->
<table width="100%" align="center" border="0" cellspacing="0" cellpadding="20">
<!-- IFOR -->
<tr>
<td width="35%">
<img src='images/simpleviewpp.gif' width=100%>
</td>
<td valign="top" width="70%">
<p>
<a href="https://ieeexplore.ieee.org/document/9874679">
<papertitle>SimpleView++: Neighborhood Views for Point Cloud Classification</papertitle>
</a>
<br>
<strong>Shivanand Venkanna Sheshappanavar</strong>, Chandra Kambhamettu
<br>
<em>MIPR 2022</em>
<br>
<a href="https://ieeexplore.ieee.org/document/9874679">[paper]</a> <a href="https://github.com/VimsLab/SimpleViewPlusPlus">[code]</a> <br>
<p align="justify">We propose the use of neighbor projections along with object projections to learn finer local structural information. SimpleView++ concatenates features from orthogonal perspective projections at object and neighbor levels with encoded features from the point cloud.</p>
</td>
</tr>
<!-- 6DOF -->
<tr>
<td width="35%">
<img src='images/patchaugment.gif' width=100%>
</td>
<td valign="top" width="70%">
<p>
<a href="https://openaccess.thecvf.com/content/ICCV2021W/DLGC/html/Sheshappanavar_PatchAugment_Local_Neighborhood_Augmentation_in_Point_Cloud_Classification_ICCVW_2021_paper.html">
<papertitle>PatchAugment: Local Neighborhood Augmentation in Point Cloud Classification</papertitle>
</a>
<br>
<strong>Shivanand Venkanna Sheshappanavar</strong>, Vinit Veerendraveer Singh, Chandra Kambhamettu
<br>
<em>ICCV Workshops 2022</em>
<br>
<a href="https://openaccess.thecvf.com/content/ICCV2021W/DLGC/papers/Sheshappanavar_PatchAugment_Local_Neighborhood_Augmentation_in_Point_Cloud_Classification_ICCVW_2021_paper.pdf">[paper]</a> <a href="https://github.com/VimsLab/PatchAugment">[code]</a> <a href="https://youtu.be/YqP7UVhwdWQ">[Video]</a> <br>
<p align="justify">Different local neighborhoods on the object surface hold a different amount of geometric complexity. Applying the same data augmentation techniques at the object level is less effective in augmenting local neighborhoods with complex structures. This paper presents PatchAugment, a data augmentation framework to apply different augmentation techniques to the local neighborhoods.</p>
</td>
</tr>
<!-- Non-deep Networks -->
<tr>
<td width="35%">
<img src='images/dynamicellipsoid.gif' width=100%>
</td>
<td valign="top" width="70%">
<p>
<a href="https://ieeexplore.ieee.org/document/9565556">
<papertitle>Dynamic local geometry capture in 3d point cloud classification</papertitle>
</a>
<br>
<strong>Shivanand Venkanna Sheshappanavar</strong>, Chandra Kambhamettu
<br>
<em>MIPR 2021</em>
<br>
<a href="https://ieeexplore.ieee.org/document/9565556">[paper]</a> <a href="https://github.com/VimsLab/DynamicScale">[code]</a> <a href="https://youtu.be/Ev44a02mwCg">[video]</a><br>
<!-- <a href="https://syncedreview.com/2021/10/22/deepmind-podracer-tpu-based-rl-frameworks-deliver-exceptional-performance-at-low-cost-129/"><img src='images/synced.png' height=25px></a>
<a href="https://papersread.ai/e/non-deep-networks/"><img src='images/podbean.png' height=25px></a> -->
<p align="justify"> PointNet++ model uses ball querying for local geometry capture in its set abstraction layers. Several models based on single scale grouping of PointNet++ continue to use ball querying with a fixed-radius ball. However, ball lacks orientation and is ineffective in capturing complex or varying geometry proportions from different local neighborhoods on the object surface. We propose a novel technique of dynamically oriented and scaled ellipsoid based on unique local information to capture the local geometry better. We also propose ReducedPointNet++, a single set abstraction based single scale grouping model. </p>
</td>
</tr>
<!-- SimpleView -->
<tr>
<td width="35%">
<img src='images/ellipsoid_querying.gif' width=100%>
</td>
<td valign="top" width="70%">
<p>
<a href="https://arxiv.org/abs/2106.05304">
<papertitle>A novel local geometry capture in pointnet++ for 3d classification</papertitle>
</a>
<br>
<strong>Shivanand Venkanna Sheshappanavar</strong>, Chandra Kambhamettu
<br>
<em>CVPR Workshops</em> 2020
<br>
<a href="https://openaccess.thecvf.com/content_CVPRW_2020/papers/w16/Sheshappanavar_A_Novel_Local_Geometry_Capture_in_PointNet_for_3D_Classification_CVPRW_2020_paper.pdf">[paper]</a>
<a href="https://github.com/VimsLab/EllipsoidQuery">[code]</a>
<a href="https://youtu.be/OMZKTH85T8c">[video]</a>
<!-- <a href="https://docs.google.com/presentation/d/1iYC5vDQPsb9nG9QCLwpm5TPL8i1F3sxu/edit?usp=sharing&ouid=116890615761823045471&rtpof=true&sd=true">[slides]</a> -->
<!-- <a href="https://icml.cc/virtual/2021/poster/9099">[video]</a> -->
<p align="justify">Few of the recent deep learning models for 3D point sets classification are dependent on how well the model captures the local geometric structures. PointNet++ model was able to extract the local region features from points by ball querying the local neighborhoods. However, ball querying is less effective in capturing local neighborhoods of high curvature surfaces or regions. In this paper, we demonstrate improvement in the 3D classification results by using ellipsoid querying around centroids, capturing more points in the local neighborhood. We extend the ellipsoid querying technique by orienting it in the direction of principal axes of the local neighborhood for better capture of the local geometry. </p>
</td>
</tr>
<!-- Rel3D -->
<tr>
<td width="35%">
<img src='images/dmc.gif' width=100%>
</td>
<td valign="top" width="70%">
<p>
<a href="https://ieeexplore.ieee.org/document/9506311">
<papertitle>Mesh Classification with Dilated Mesh Convolutions</papertitle>
</a>
<br>
Vinit Veerendraveer Singh, <strong>Shivanand Venkanna Sheshappanavar</strong>, Chandra Kambhamettu
<br>
<em>ICIP</em> 2021
<br>
<!-- <em>NeuRIPS</em> 2020, <span style="color:brown;">Spotlight (Top 4% of submitted papers)</span>
<br> -->
<a href="https://ieeexplore.ieee.org/document/9506311">[paper]</a>
<a href="https://github.com/VimsLab/DMC">[code]</a>
<!-- <a href="https://drive.google.com/file/d/1lWA2WlJR4itJJrnHMRiZ87-z8akagkH4/view?usp=sharing">[slides]</a> -->
<a href="https://youtu.be/Jdl71d3oMRE">[video]</a>
<p align="justify"> In this paper, inspired by dilated convolutions for images, we proffer dilated convolutions for meshes. Our Dilated Mesh Convolution (DMC) unit inflates the kernels' receptive field without increasing the number of learnable parameters. We also propose a Stacked Dilated Mesh Convolution (SDMC) block by stacking DMC units. We accommodated SDMC in MeshNet to classify 3D meshes. </p>
</td>
</tr>
<!-- PackIt -->
<tr>
<td width="35%">
<img src='images/meshnet2.gif' width=100%>
</td>
<td valign="top" width="70%">
<p>
<a href="https://arxiv.org/abs/2007.11121">
<papertitle>MeshNet++: A Network with a Face</papertitle>
</a>
<br>
Vinit Veerendraveer Singh, <strong>Shivanand Venkanna Sheshappanavar</strong>, Chandra Kambhamettu
<br>
<em>ACM MM</em> 2021
<br>
<a href="https://dl.acm.org/doi/pdf/10.1145/3474085.3475468">[paper]</a>
<a href="https://github.com/VimsLab/MeshNet2">[code]</a>
<a href="https://youtu.be/xcfnhrYqKac">[video]</a>
<!-- <a href="data/packit_slides.pptx">[slides]</a> -->
<p align="justify"> MeshNet is a pioneer in this direction. In this paper, we propose a novel neural network that is substantially deeper than its MeshNet predecessor. This increase in depth is achieved through our specialized convolution and pooling blocks that operate on mesh faces. Our network named MeshNet++ learns local structures at multiple scales and is also robust to shortcomings of mesh decimation. We evaluated it for the shape classification task on various data sets, and results significantly higher than state-of-the-art were observed.</p>
</td>
</tr>
<tr>
<td width="35%">
<img src='images/soil_moisture.gif' width=100%>
</td>
<td valign="top" width="70%">
<p>
<a href="https://arxiv.org/abs/2007.11121">
<papertitle>LSTM based Soil Moisture Prediction</papertitle>
</a>
<br>
<strong>Shivanand Venkanna Sheshappanavar</strong>, Chilukuri K. Mohan, David G. Chandler
<br>
<em>NERCS</em> 2018
<br>
<a href="https://github.com/sheshap/sheshap.github.io/blob/master/pdf/SoilMoisturePrediction_LSTM_NERCCS_2018.pdf">[paper]</a>
<a href="https://github.com/sheshap/SoilMoisturePrediction">[code]</a>
<!-- <a href="https://youtu.be/xcfnhrYqKac">[video]</a> -->
<!-- <a href="data/packit_slides.pptx">[slides]</a> -->
<p align="justify"> Soil moisture content is an important variable that has a considerable impact on agricultural processes and practical weather-related concerns such as flooding and drought. We address the problem of predicting soil moisture by applying recurrent neural networks that use Long Short-Term Memory (LSTM) models. The success of our approach is evaluated using a dataset obtained from ground-based sensor infrastructure networks. Feature reduction using a mutual information approach is shown to be more effective than feature extraction using principal component analysis.</p>
</td>
</tr>
</table>
<!-- Teaching -->
<table width="100%" align="center" border="0" cellspacing="0" cellpadding="20">
<tr>
<td width="100%" valign="middle">
<heading>Teaching</heading>
<p>
I have been the Instructor for the following course:
<ul>
<li>CISC210: Introduction to Systems Programming [Summer 2020]</li>
</ul>
I have been the Lead Teaching Assistant for the following course:
<ul>
<li> CISC210: Introduction to Systems Programming at the University of Delaware[Fall 2022, Spring 2022, Spring 2021, Fall 2020, Spring 2020, Fall 2019, Spring 2019] </li>
</ul>
I have been the Teaching Assistant for the following courses:
<ul>
<li> CISC220: Data Structures at the University of Delaware [Fall 2021] </li>
<li> CISC101: Principles of Computing at the University of Delaware [Winter 2021] </li>
<li> CISC662: Advanced Computer Architecture at the University of Delaware [Fall 2018]</li>
</ul>
</p>
<!-- Some <a href="https:/related.html">related papers</a> to mine. -->
</td>
</tr>
</table>
<!-- Reference -->
<p align="right"><a href="https://jonbarron.info/">[Web Cite]</a></p>
</td>
</tr>
