initial commit

_config.yml

@@ -7,13 +7,13 @@ url: 'https://cvlab-unibo.github.io'
 baseurl:
 
 # Title of the Site
-title: Learning Depth Estimation for Transparent and Mirror Surfaces
+title: Boosting Multi-Modal Unsupervised Domain Adaptation for LiDAR Semantic Segmentation by Self-Supervised Depth Completion
 # Description of the Site
-description: Accepted at ICCV2023
+description: Accepted at IEEE Access
 # URL of Image of the Site
 site-image:
 # Keywords of the Site
-site-keywords: Deep Learning, Depth, Disparity, Segmentation, Monocular, Mono, Stereo, Transparent, Glass, Specular, Reflective, non-Lambertian
+site-keywords: Deep Learning, Depth, Disparity, Segmentation, Domain Adaptation, Lidar
 # URL for the Image of custom Favicon
 favicon-url:
 
@@ -27,7 +27,7 @@ facebook_username:
 twitter_username:
 instagram_username:
 linkedin_username:
-github_username:  arcanoXIII/learning_depth_estimation_for_transparent_and_mirror_surfaces
+github_username:  pierlui92/cts-web
 youtube_channel_id: 
 reddit_username:
 behance_username:
@@ -39,10 +39,10 @@ site_name: "https://www.vision.deis.unibo.it/"
 google-analytics:
 
 # Disqus
-disqus-shortname: Depth4ToM
+disqus-shortname: CtS
 
 # Name of the Author
-author-name: Alex Costanzino
+author-name: Pierluigi Zama Ramirez
 # URL for the Image of the Author
 author-image:
 # 60 Words About the Author

_includes/footer.html

@@ -5,6 +5,6 @@
         <a class="has-text-white" onclick="window.scroll(0,0)">BACK TO TOP</a>
     </div>
     <div class="has-background-black has-text-centered has-text-white" id="credits">
-        <i class="far fa-copyright"></i> {{ "now" | date: "%Y" }} | <a target="_blank" rel="noopener noreferrer"> Depth4ToM </a> - A site powered by <a href="https://jekyllrb.com/" target="_blank">Jekyll</a> a fork of <a href="https://github.com/thedevslot/WhatATheme" target="_blank">WhatATheme</a> 
+        <i class="far fa-copyright"></i> {{ "now" | date: "%Y" }} | <a target="_blank" rel="noopener noreferrer"> CtS </a> - A site powered by <a href="https://jekyllrb.com/" target="_blank">Jekyll</a> a fork of <a href="https://github.com/thedevslot/WhatATheme" target="_blank">WhatATheme</a> 
     </div>
 </footer>

_includes/home.html

@@ -3,11 +3,7 @@
       <div class="container">
         <h2 align="centered" class="title"> Abstract </h2>
         <p style = "text-align:justify ; text-justify:inter-word">
-            Inferring the depth of transparent or mirror (ToM) surfaces represents a hard challenge for either sensors, algorithms, or deep networks. 
-            We propose a simple pipeline for learning to estimate depth properly for such surfaces with neural networks, without requiring any ground-truth annotation. 
-            We unveil how to obtain reliable pseudo labels by in-painting ToM objects in images and processing them with a monocular depth estimation model. 
-            These labels can be used to fine-tune existing monocular or stereo networks, to let them learn how to deal with ToM surfaces. 
-            Experimental results on the Booster dataset show the dramatic improvements enabled by our remarkably simple proposal.
+
         </p>
     </div>
 </div>

_includes/paper.html

@@ -1,35 +1,28 @@
 <section class="hero  has-text-centered" id="paper">
     <div class="hero-body">
         <div class="container">
-          <h2 align="centered" class="title">Paper Depth4ToM</h2>
+          <h2 align="centered" class="title">PAPER</h2>
             <div class="columns">
                 <div class="column is-one-fifth-desktop is-one-fifth-tablet is-one-fifth-fullhd">
                   <br>
                   <br>
                   <br>
-                  <a href="https://arxiv.org/abs/2307.15052">
-                  <img style="width:90%" src="assets/paper_icon.png" >
+                  <a href="">
+                  <img style="width:90%" src="assets/images/paper_image.jpg" >
                   </a>
                 </div>
                 <div class="column has-text-left-desktop has-text-left-tablet has-text-left-fullhd has-text-left-widescreen">
 
                     <br>
                     <div class="container columns is-centered">
                         <div>
-                          <B><a href="https://arxiv.org/abs/2307.15052"><font size = "+2">Learning Depth Estimation for Transparent and Mirror Surfaces</font></a><br></B>
-                          <B><i> Alex Costanzino*, Pierluigi Zama Ramirez*, Matteo Poggi*, Fabio Tosi, Stefano Mattoccia, Luigi Di Stefano</B></i>
-                          <br>
-                          <i><font size = "-1">*Equal Contribution</font></i>
+                          <B><a href=""><font size = "+2">Boosting Multi-Modal Unsupervised Domain Adaptation for LiDAR Semantic Segmentation by Self-Supervised Depth Completion</font></a><br></B>
+                          <B><i> Adriano Cardace, Andrea Conti, Pierluigi Zama Ramirez, Riccardo Spezialetti, Samuele Salti, Luigi Di Stefano</B></i>
                           <br>
                           <br>
                           <p>
-                            Inferring the depth of transparent or mirror (ToM) surfaces represents a hard challenge for either sensors, algorithms, or deep networks. 
-
-                            We propose a simple pipeline for learning to estimate depth properly for such surfaces with neural networks, without requiring any ground-truth annotation. 
-                            We unveil how to obtain reliable pseudo labels by in-painting ToM objects in images and processing them with a monocular depth estimation model. 
-                            These labels can be used to fine-tune existing monocular or stereo networks, to let them learn how to deal with ToM surfaces. 
-
-                            Experimental results on the Booster dataset show the dramatic improvements enabled by our remarkably simple proposal.
+                            LiDAR semantic segmentation is receiving increased attention due to its deployment in autonomous driving applications. As LiDARs come often with other sensors such as RGB cameras, multi-modal approaches for this task have been developed, which however suffer from the domain shift problem as other deep learning approaches. To address this, we propose a novel Unsupervised Domain Adaptation (UDA) technique for multi-modal LiDAR segmentation. Unlike previous works in this field, we leverage depth completion as an auxiliary task to align features extracted from 2D images across domains, and as a powerful data augmentation for LiDARs.
+                            We validate our method on three popular multi-modal UDA benchmarks and we achieve better performances than other competitors.
                           </div>
                         <div>
 
@@ -44,12 +37,14 @@ <h3>
                 </h3>
                 <div class="has-text-left-desktop has-text-left-tablet has-text-left-fullhd has-text-left-widescreen form-group col-md-18 col-md-offset-0">
 <pre>
-@inproceedings{costanzino2023iccv,
-    title = {Learning Depth Estimation for Transparent and Mirror Surfaces},
-    author = {Costanzino, Alex and Zama Ramirez, Pierluigi and Poggi, Matteo and Tosi, Fabio and Mattoccia, Stefano and Di Stefano, Luigi},
-    booktitle = {The IEEE International Conference on Computer Vision},
-    note = {ICCV},
-    year = {2023},
-}
+
+  @article{cardace2023cts,
+    title={Boosting Multi-Modal Unsupervised Domain Adaptation for LiDAR Semantic Segmentation by Self-Supervised Depth Completion},
+    author={Cardace, Adriano and Conti, Andrea and Zama Ramirez, Pierluigi and Spezialetti, Riccardo and Salti, Samuele and Di Stefano, Luigi},
+    journal={IEEE Access},
+    year={2023},
+    publisher={IEEE}
+  }
+
 </pre>
 </section>

_includes/us.html

@@ -76,11 +76,18 @@ <h1>US</h1>
 
 
       <div class="pe">
-        <img src="assets/images/alex.jpg" alt="">
-        <div class="p-name">Alex Costanzino</div>
+        <img src="assets/images/adri.png" alt="">
+        <div class="p-name">Adriano Cardace</div>
         <div class="p-des">PhD Student</div>
         <div class="p-des">University of Bologna</div>
-        <div class="p-des">[email protected]</div>
+        <div class="p-des">[email protected]</div>
+      </div>
+      <div class="pe">
+        <img src="assets/images/Andre.jpg" alt="">
+        <div class="p-name">Andrea Conti</div>
+        <div class="p-des">PhD Student</div>
+        <div class="p-des">University of Bologna</div>
+        <div class="p-des">[email protected]</div>
       </div>
       <div class="pe">
         <img src="assets/images/pier.jfif" alt="">
@@ -90,25 +97,18 @@ <h1>US</h1>
         <div class="p-des">[email protected]</div>
       </div>
       <div class="pe">
-        <img src="assets/images/matteo.jpeg" alt="">
-        <div class="p-name">Matteo Poggi</div>
-        <div class="p-des">Assistant Professor</div>
-        <div class="p-des">University of Bologna</div>
-        <div class="p-des">[email protected]</div>
-      </div>
-      <div class="pe">
-        <img src="assets/images/fabio.png" alt="">
-        <div class="p-name">Fabio Tosi</div>
+        <img src="assets/images/marcello.jpg" alt="">
+        <div class="p-name">Riccardo Spezialetti</div>
         <div class="p-des">Post Doc</div>
         <div class="p-des">University of Bologna</div>
-        <div class="p-des">fabio.tosi5@unibo.it</div>
+        <div class="p-des">riccardo.spezialti@unibo.it</div>
       </div>
       <div class="pe">
-        <img src="assets/images/stefano.jpg" alt="">
-        <div class="p-name">Stefano Mattoccia</div>
+        <img src="assets/images/Samuele.jpg" alt="">
+        <div class="p-name">Samuele Salti</div>
         <div class="p-des">Professor</div>
         <div class="p-des">University of Bologna</div>
-        <div class="p-des">stefano.mattoccia@unibo.it</div>
+        <div class="p-des">samuele.salti@unibo.it</div>
       </div>
       <div class="pe">
         <img src="assets/images/luigi.jpg" alt="">