Skip to content

Commit

Permalink
add horizongs
Browse files Browse the repository at this point in the history
  • Loading branch information
@jianglh-WHU
jianglh-WHU committed Nov 27, 2024
1 parent d19bc09 commit cc59b9d
Show file tree
Hide file tree
Showing 27 changed files with 14,804 additions and 1 deletion.
335 changes: 335 additions & 0 deletions horizon-gs/index.html
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,335 @@
<!DOCTYPE html lang="en">
<html>
<head>
<meta charset="utf-8">
<meta name="description" content="Horizon-GS: Unified 3D Gaussian Splatting for Large-Scale Aerial-to-Ground Scenes">
<meta name="keywords" content="Horizon-GS">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Horizon-GS: Unified 3D Gaussian Splatting for Large-Scale Aerial-to-Ground Scenes</title>

<!-- Bootstrap -->
<link rel="stylesheet" href="./static/css/bootstrap-4.4.1.css">

<!-- Global site tag (gtag.js) - Google Analytics -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-PYVRSFMDRL"></script>
<script>
window.dataLayer = window.dataLayer || [];

function gtag() {
dataLayer.push(arguments);
}

gtag('js', new Date());

gtag('config', 'G-PYVRSFMDRL');
</script>

<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro" rel="stylesheet">
<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
<link rel="stylesheet" href="./static/css/bulma-slider.min.css">
<link rel="stylesheet" href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<link rel="icon" href="./static/images/icon.png">

<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="./static/js/bulma-carousel.min.js"></script>
<script src="./static/js/bulma-slider.min.js"></script>
<script src="./static/js/index.js"></script>

<link rel="stylesheet" href="./static/css/dics.original.css">
<script src="./static/js/event_handler.js"></script>
<script src="./static/js/dics.original.js"></script>
<script src="./static/js/video-comparison.js"></script>

<style>
.container {
width: 80%;
margin: auto;
}

.responsive-video {
width: 45%;
height: auto;
}
</style>

<style>
.fixed-size-video {
width: 940px;
height: auto;
}
</style>

<style>
.fixed-size-video-small {
width: 700px;
height: auto;
}
.rounded-corners {
border-radius: 10px;
}
</style>
<style>
.video-container {
height: 600px;
width: auto;
}
</style>
<style>
.video-container {
display: flex;
justify-content: center;
}
.video-container video {
width: 30%;
margin: 5px;
}
</style>

<style>
.video-row {
display: flex;
justify-content: space-around;
margin-bottom: 20px;
}

.fixed-size-video-small {
width: 33%;
height: auto;
}
.fixed-size-video-small-two {
width: 800px;
height: auto;
}
.fixed-size-video-small-three {
width: 1340px;
height: auto;
}

@media screen and (max-width: 768px) {
.video-row {
flex-direction: column;
}
}
</style>

<style>
.text-container {
max-width: 1070px;
margin: 0 auto;
padding: 20px;
}

</style>
</head>
<body>

<section class="hero">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-2 publication-title">Horizon-GS: Unified 3D Gaussian Splatting for Large-Scale Aerial-to-Ground Scenes</h1>
<div class="is-size-5 publication-authors">
<span class="author-block"><a href="https://jianglh-whu.github.io/">Lihan Jiang*</a><sup>1,3</sup>,</span>
<span class="author-block"><a href="https://github.com/tongji-rkr">Kerui Ren*</a><sup>2,3</sup>,</span>
<span class="author-block"><a href="https://scholar.google.com/citations?user=w0Od3hQAAAAJ">Mulin Yu</a><sup>3</sup>,</span>
<span class="author-block"><a href="https://eveneveno.github.io/lnxu">Linning Xu</a><sup>4</sup>,</span> <br>
<span class="author-block"><a href="https://jtdong.com/">Junting Dong</a><sup>3</sup>,</span>
<span class="author-block"><a href="https://inspirelt.github.io/">Tao Lu</a><sup>5</sup>,</span>
<span class="author-block"><a href="https://scholar.google.co.uk/citations?user=r6CvuOUAAAAJ&hl=en">Feng Zhao</a><sup>1</sup>,</span>
<span class="author-block"><a href="http://dahua.site/">Dahua Lin</a><sup>4</sup>,</span>
<span class="author-block"><a href="https://daibo.info/">Bo Dai</a><sup>6,3</sup></span>
</div>

<div class="is-size-5 publication-authors">
<span class="author-block"><sup>1</sup> University of Science and Technology of China </span>,
<span class="author-block"><sup>2</sup> Shanghai Jiao Tong University</span>,
<span class="author-block"><sup>3</sup> Shanghai Artificial Intelligence Laboratory </span>,
<span class="author-block"><sup>4</sup> The Chinese University of Hong Kong </span> <br>
<span class="author-block"><sup>5</sup> Brown University </span>
<span class="author-block"><sup>6</sup> The University of Hong Kong </span>
</div>

<div class="column has-text-centered">
<div class="publication-links">
<!-- PDF Link. -->
<span class="link-block">
<a href=""
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<!-- Code Link. -->
<span class="link-block">
<a href=""
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>Code (Coming Soon)</span>
</a>
</span>
</div>
</div>
</div>
</div>
</div>
</div>
</section>

<section class="hero teaser">

<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<div class="text-container">
<p style="font-size:22px;">
TL;DR: We introduce <b>Horizon</b>-GS, tackles the unified reconstruction and rendering for aerial and street views with a new training strategy, overcoming viewpoint discrepancies to generate high-fidelity scenes.
<!-- Our model dynamically selects the appropriate level from the set of multi-resolution anchor points, ensuring <b>consistent</b> rendering performance with adaptive LOD adjustments while maintaining <b>high-fidelity</b> rendering. -->
</p>
</div>
<video class="fixed-size-video rounded-corners" id="supp video" autoplay loop muted playsinline controls>
<source src="./static/images/demo.mp4" type="video/mp4">
</video>

<div class="text-container">
<p style="font-size:22px; width: 95%; margin: auto;">
Our method delivers an immersive and seamless experience for city roaming while achieving high-quality rendering and reconstruction of aerial-to-ground scenes.
</p>
</div>

<!-- <br><br> -->

<div class="content has-text-centered" style="width: 80%; margin: auto;">
<img src="./static/images/Teaser.png">
</div>

<div class="text-container">
<p style="font-size:22px; width: 95%; margin: auto;">
The colored camera trajectories depict novel viewpoints, with the reconstructed mesh overlaid on the scene, while the surrounding images display the predicted views for each trajectory.
</p>
</div>
</div>
</div>

<br><br>

<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<div class="hero-body">
<h2 class="title is-3">Abstract</h2>
<div class="content has-text-justified">
<p style="font-size:18px;">
Seamless integration of both aerial and street view images remains a significant challenge in neural scene reconstruction and rendering.
Existing methods predominantly focus on single domain, limiting their applications in immersive environments, which demand extensive free view exploration with large view changes both horizontally and vertically.
We introduce Horizon-GS, a novel approach built upon Gaussian Splatting techniques, tackles the unified reconstruction and rendering for aerial and street views.
Our method addresses the key challenges of combining these perspectives with a new training strategy, overcoming viewpoint discrepancies to generate high-fidelity scenes.
We also curate a high-quality aerial-to-ground views dataset encompassing both synthetic and real-world scene to advance further research.
Experiments across diverse urban scene datasets confirm the effectiveness of our method.
</p>
</div>
</div>
</div>
</div>
</div>


<br><br>

<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-full-width">
<div class="hero-body">
<h2 class="title is-3">Method Overview</h2>
<img src="./static/images/pipeline.png">
<div class="content has-text-justified">
<p style="margin-top: 30px; font-size:18px;">
<b> Illustration of our proposed Horizon-GS: </b> We divide large-scale scenes into chunks. For each chunk,
we initialize LOD-structured anchors and conduct the coarse-to-fine training process. Specifically, the coarse stage reconstructs the overall scene,
while the fine stage enhances street view details (highlighted in <span style="color: rgb(192, 113, 255);">purple</span>). We can derive RGB, depth,
and normal images by utilizing different primitive attributes (2D/3D Gaussians) with a single shared underlying structure.
</p>
</div>
</div>
</div>
</div>
</div>

<br><br>

<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-full-width">
<div class="hero-body">
<h2 class="title is-3">Data Capture</h2>
<img src="./static/images/data.png">
<div class="content has-text-justified">
<p style="margin-top: 30px; font-size:18px;">
<b> Visualization of our constructed dataset: </b> All the 7 scenes contain calibrated aerial and street view images.
We illustrate the scenes with the point clouds and the corresponding image capture poses. The trajectory of aerial views is shown in <span style="color: rgb(192, 113, 255);">purple</span>,
while street views are represented in <span style="color: rgb(250, 176, 26);">yellow</span>. Our dataset contains 5 synthetic scenes (a-e) and 2 real scenes (f-g).
</p>
</div>
</div>
</div>
</div>
</div>

<!-- <br><br> -->

</section>

<section class="section">
<div class="columns is-centered ">
<div class="column is-full-width">
<h3 class="title is-3 has-text-centered" style="margin-bottom: 40px">Results</h2>
</div>
</div>

<div class="hero-body" style="margin-top: -60px;">
<div class="container is-max-desktop" style="text-align:left">
<h3 class="title is-4">Rendering Performance</h3>
<p style="font-size:18px; font-weight: 300;">
Compared to baselines, <em>Horizon-GS</em> successfully captures fine details in the scene, particularly for objects with thin structures such as trees, decorative texts, etc, from delicate scenes (a) to large-scale scenes (b).
</p>
<br>
<div class="content has-text-centered" style="width: 100%; margin: auto;">
<img src="./static/images/main.png">
</div>
</div>
<div class="container is-max-desktop" style="text-align:left; margin-top: 30px;">
<h3 class="title is-4">Surface Performance</h3>
<p style="font-size:18px; font-weight: 300;">
Thanks to the two-stage training approach, <em>Horizon-GS</em> can delivers geometrically accurate, and artifact-free reconstruction. In contrast, 2D-GS introduces artifacts, resulting in incomplete and lackluster geometry.
</p>
<br>
<div class="content has-text-centered" style="width: 100%; margin: auto;">
<img src="./static/images/surface.png">
</div>
</div>
</section>


<footer class="footer">
<div class="container">
<div class="columns is-centered">
<div class="column is-8">
<div class="content">
<p>
We thank the authors of <a href="https://nerfies.github.io/">Nerfies</a> that kindly open sourced the template of this website.
</p>
</p>
</div>
</div>
</div>
</div>
</footer>

</body>
</html>
Loading

0 comments on commit cc59b9d

Please sign in to comment.