Update

index.html

@@ -10,9 +10,9 @@
 
 <div class="header">
     <h2 class="header-content" id="title">HILO: A Large-Scale Heterogeneous Object Dataset for Benchmarking Robotic Grasping Approaches<br></h2>
-    <h3 class="header-content" id="conference"><i>Published in ICARA 2025</i></h3>
+    <h3 class="header-content" id="conference"><i>Accepted and Presented at ICARA 2025</i></h3>
     <p class="header-content" id="authors">
-        Xinchao Song, Sean Banerjee, and Natasha Kholgade Banerjee
+        Xinchao Song, Sean Banerjee, Natasha Kholgade Banerjee
     </p>
     <p class="header-content" id="authors">
         Terascale All sensing Research Studio
@@ -24,7 +24,6 @@ <h3 class="header-content" id="conference"><i>Published in ICARA 2025</i></h3>
                 src="assets/drive_icon.png" alt="Google drive icon" class="icon">Dataset</a>
         <a href="https://github.com/Terascale-All-sensing-Research-Studio/hilo_dataset" target="_blank" class="link button"><img
                 src="assets/github_mark.png" alt="GitHub icon" class="icon">Code</a>
-
     </p>
 </div>
 
@@ -34,20 +33,22 @@ <h3 class="header-content" id="conference"><i>Published in ICARA 2025</i></h3>
         <br>
         <h3 class="body-title">Abstract</h3>
         <p id="abstract-body">
-            Robust object manipulation is essential for robotics applications in real-world environments, especially when handling diverse and complex everyday objects. To facilitate this research, we present HILO, a large-scale dataset of 253 everyday objects and 288 diverse scenes. HILO bridges a crucial gap in existing manipulation datasets through its heterogeneity and dual-resolution approach, combining high-resolution individual object scans with low-resolution scans of cluttered scenes. This provides both the precise geometric data needed for grasp planning and realistic environmental context. The dataset's comprehensive representations enable rigorous benchmarking of robotic grasping algorithms. Our evaluation of three leading grasping algorithms—Contact-GraspNet, GraspNet Baseline, and DexNet 4.0—reveals critical trade-offs between grasp quantity and quality, demonstrating the dataset's value in advancing robotic grasping research. HILO's rich object diversity and dual-resolution methodology provide a foundation for developing more versatile robotic systems capable of reliable real-world robotic manipulation.
+            Robust object manipulation is essential for robotics applications in real-world environments, especially when handling diverse and complex everyday objects. To facilitate this research, we present <b>HILO</b>, a large-scale dataset of 253 everyday objects and 288 diverse scenes. HILO bridges a crucial gap in existing manipulation datasets through its heterogeneity and dual-resolution approach, combining <b>HI</b>gh-resolution individual object scans with <b>LO</b>w-resolution scans of cluttered scenes. This provides both the precise geometric data needed for grasp planning and realistic environmental context. The dataset's comprehensive representations enable rigorous benchmarking of robotic grasping algorithms. Our evaluation of three leading grasping algorithms—Contact-GraspNet, GraspNet Baseline, and DexNet 4.0—reveals critical trade-offs between grasp quantity and quality, demonstrating the dataset's value in advancing robotic grasping research. HILO's rich object diversity and dual-resolution methodology provide a foundation for developing more versatile robotic systems capable of reliable real-world robotic manipulation.
         </p>
     </div>
 
     <br>
 
-    <h3 class="body-title">Dataset Objects</h3>
+    <h3 class="body-title"><b>HI</b>gh-Resolution Models</h3>
 
     <figure>
-        <img src="assets/fig_hilo_objects.png"
-             alt="Example objects from the HILO dataset." id="fig_hilo_objects"
-             height="100%" width="100%">
+        <img src="assets/fig_hilo_objects.jpg"
+             class="img-fluid"
+             alt="Example high-resolution models from the HILO dataset."
+             id="fig_hilo_objects"
+             width="640">
         <figcaption>
-            The HILO dataset comprises 253 diverse everyday objects across seven categories: toys, food and drink items, cooking utensils, tools, mugs and containers, general household items, and office supplies. All these items are easily obtainable through retail or online vendors. 
+            The HILO dataset comprises the high-resolution (HI) individual scanned models for 253 diverse everyday objects across seven categories: toys, food and drink items, cooking utensils, tools, mugs and containers, general household items, and office supplies. All these items are easily obtainable through retail or online vendors. 
         </figcaption>
     </figure>
 
@@ -57,16 +58,16 @@ <h3 class="body-title">Dataset Objects</h3>
         <table style="margin-left: auto;  margin-right: auto; border-collapse: collapse">
             <caption style="margin-top: 15px; caption-side: top; line-height: 150%; text-align: center;
             color: black">
-                The number of objects, average mass, vertex count, face count, volume, and surface area per category of the HILO dataset.
+                The number of objects, average mass, vertex count, face count, volume, and surface area per category of the HILO dataset
             </caption>
             <tr class="border-vertical1">
                 <th></th>
                 <th class="border-left">#Objects</th>
                 <th>Mass (<i>g</i>)</th>
                 <th>Vertices</th>
                 <th>Faces</th>
-                <th>Volume (<i>{cm}^3$)</th>
-                <th>Surface Area (<i>{cm}^2</i>)</th>
+                <th>Volume (<i>cm<sup>3</sup></i>)</th>
+                <th>Surface Area (<i>cm<sup>2</sup></i>)</th>
             </tr>
             <tr class="border-bottom">
                 <td>Toys</td>
@@ -146,35 +147,42 @@ <h3 class="body-title">Dataset Objects</h3>
     <br>
     <br>
 
-    <h3 class="body-title">Dataset Scenes</h3>
+    <h3 class="body-title"><b>LO</b>w-Resolution Scenes</h3>
 
     <div class="plain-text">
-    The HILO dataset contains 288 total scenes. Each scene contains:
+    The HILO dataset contains 288 low-resolution cluttered scenes from 72 groups of 10 objects randomly selected from the 253 objects. Each scene contains:
 
     <ul>
-        <li>120 or 104 raw RGBD images for 10 different objects</li>
+        <li>120 or 104 raw RGBD images from different viewpoints</li>
         <li>120 or 104 corresponding undistorted RGBD images</li>
         <li>Masks for the undistorted RGBD images</li>
         <li>Object annotations</li>
     </ul>
 
+    <br>
+
     </div>
 
     <figure>
-        <img src="assets/hilo_scenes.png"
-             alt="HILO RGB image examples"
-             id="fig_hilo_scene"
-             height="100%" width="100%">
-        <figcaption>The HILO dataset contains 32,256 diverse RGBD image grouped into 288 scenes.
+        <img src="assets/hilo_example_scenes.jpg"
+             class="img-fluid"
+             alt="Example low-resolution scenes from the HILO dataset."
+             id="hilo_example_scenes"
+             width="640">
+        <figcaption>The HILO dataset contains 32,256 RGBD image from diverse viewpoints.
         </figcaption>
     </figure>
 
+    <br>
+    <br>
+
     <figure>
-        <img src="assets/fig_hilo_aligned_scene.png"
-             alt="An example point cloud generated by the undistorted RGBD image and aligned with high-resolution meshes using the corresponding pose transformation annotation."
-             id="fig_hilo_scene"
-             height="100%" width="100%">
-        <figcaption>An example point cloud generated by the undistorted RGBD image and aligned with high-resolution meshes using the corresponding object transformation annotations.
+        <img src="assets/hilo_scene_pcd.gif"
+             class="img-fluid"
+             alt="A GIT animation of an example point cloud generated by the masked undistorted RGBD image of a low-resolution scene and aligned with each high-resolution mesh."
+             id="hilo_scene_pcd"
+             width="640">
+        <figcaption>An example point cloud generated by the masked undistorted RGBD image of a low-resolution scene and aligned with high-resolution meshes using the corresponding object transformation annotations.
         </figcaption>
     </figure>