semissc.github.io/index.html at master · semissc/semissc.github.io · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
<!DOCTYPE html>
<html>
<head>
  <meta charset="utf-8">
  <!-- Meta tags for social media banners, these should be filled in appropriatly as they are your "business card" -->
  <!-- Replace the content tag with appropriate information -->
  <meta name="description" content="DESCRIPTION META TAG">
  <meta property="og:title" content="SOCIAL MEDIA TITLE TAG"/>
  <meta property="og:description" content="SOCIAL MEDIA DESCRIPTION TAG TAG"/>
  <meta property="og:url" content="URL OF THE WEBSITE"/>
  <!-- Path to banner image, should be in the path listed below. Optimal dimenssions are 1200X630-->
  <meta property="og:image" content="static/image/your_banner_image.png" />
  <meta property="og:image:width" content="1200"/>
  <meta property="og:image:height" content="630"/>


  <meta name="twitter:title" content="TWITTER BANNER TITLE META TAG">
  <meta name="twitter:description" content="TWITTER BANNER DESCRIPTION META TAG">
  <!-- Path to banner image, should be in the path listed below. Optimal dimenssions are 1200X600-->
  <meta name="twitter:image" content="static/images/your_twitter_banner_image.png">
  <meta name="twitter:card" content="summary_large_image">
  <!-- Keywords for your paper to be indexed by-->
  <meta name="keywords" content="semantic, scene, completion, aaai, monocular, vision-based">
  <meta name="viewport" content="width=device-width, initial-scale=1">


  <title>VFG-SSC: Semi-supervised 3D Semantic Scene Completion with 2D Vision Foundation Model Guidance</title>
  <!-- <link rel="icon" type="image/x-icon" href="static/images/favicon.ico"> -->
  <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="static/css/fontawesome.all.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">

  <script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
  <script src="https://documentcloud.adobe.com/view-sdk/main.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>

  <!--MathJax-->
  <script>
    window.MathJax = {
      tex: {
        inlineMath: [['$', '$'], ['\\(', '\\)']]
      },
      svg: {
        fontCache: 'global'
      }
    };
  </script>
  <script type="text/javascript" id="MathJax-script" async src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js"></script>

</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-1 publication-title">VFG-SSC: Semi-supervised 3D Semantic Scene Completion with 2D Vision Foundation Model Guidance</h1>
            <h3 class="title has-text-centered">The 39th Annual AAAI Conference on Artificial Intelligence</h3>
            <div class="is-size-5 publication-authors">
              <!-- Paper authors -->
              <span class="author-block">
                <a href="https://haiphamcse.github.io" target="_blank">Duc-Hai Pham</a><sup>1</sup>,</span>
                <span class="author-block">
                <a href="https://scholar.google.com/citations?user=xV7uHJgAAAAJ&hl=en" target="_blank">Duc-Dung Nguyen</a><sup>2</sup>,</span>
                <span class="author-block">
                <a href="https://scholar.google.com/citations?user=sGPFOAYAAAAJ&hl=en" target="_blank">Anh Pham</a><sup>2</sup>,</span>
                <span class="author-block">
                <a href="THIRD AUTHOR PERSONAL LINK" target="_blank">Tuan Ho</a><sup>1</sup>,</span>
                <span class="author-block">
                <a href="https://phongnhhn.info" target="_blank">Phong Nguyen</a><sup>1</sup>,</span>
                <span class="author-block">
                <a href="https://khoinguyen.org" target="_blank">Khoi Nguyen</a><sup>1</sup>,</span>
                <span class="author-block">
                <a href="https://rangnguyen.github.io" target="_blank">Rang Nguyen</a><sup>1</sup></span>

                </span>
                  </div>

                  <div class="is-size-5 publication-authors">
                    <span class="author-block"> <sup>1</sup> VinAI Research, Vietnam <sup>2</sup> AITech Lab., Ho Chi Minh City University of Technology, VNU-HCM, Vietnam </span>
                  </div>

                  <div class="column has-text-centered">
                    <div class="publication-links">
                         <!-- Arxiv PDF link -->
                      <span class="link-block">
                        <a href="https://arxiv.org/pdf/2408.11559" target="_blank"
                        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>

                  <!-- Github link -->
                  <span class="link-block">
                    <a href="https://github.com/YOUR REPO HERE" target="_blank"
                    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>

                <!-- ArXiv abstract Link -->
                <span class="link-block">
                  <a href="https://arxiv.org/abs/2408.11559" target="_blank"
                  class="external-link button is-normal is-rounded is-dark">
                  <span class="icon">
                    <i class="ai ai-arxiv"></i>
                  </span>
                  <span>arXiv</span>
                </a>
              </span>
            </div>
          </div>
        </div>
      </div>
    </div>
  </div>
</section>


<!-- Teaser video-->
<section class="hero teaser">
  <div class="container is-max-desktop">
    <div class="hero-body">
      <video poster="" id="tree" autoplay controls muted loop height="100%">
        <!-- Your video here -->
        <source src="static/images/semissc_after.mp4"
        type="video/mp4">
      </video>
      <h2 class="subtitle has-text-centered">
        Vision-based 3D semantic scene completion is a crucial 3D scene understanding task. Can we train these models with limited 3D supervision?
      </h2>
    </div>
  </div>
</section>
<!-- End teaser video -->

<!-- Paper abstract -->
<section class="section hero is-light">
  <div class="container is-max-desktop">
    <div class="columns is-centered has-text-centered">
      <div class="column is-four-fifths">
        <h2 class="title is-3">Abstract</h2>
        <div class="content has-text-justified">
          <p>
            Accurate prediction of 3D semantic occupancy from 2D visual images is vital in enabling autonomous agents to comprehend their surroundings for planning and navigation. State-of-the-art methods typically employ fully supervised approaches, necessitating a huge labeled dataset acquired through expensive LiDAR sensors and meticulous voxel-wise labeling by human annotators. The resource-intensive nature of this annotating process significantly hampers the application and scalability of these methods. We introduce a novel semi-supervised framework to alleviate the dependency on densely annotated data. Our approach leverages 2D foundation models to generate essential 3D scene geometric and semantic cues, facilitating a more efficient training process. Our framework exhibits notable properties: (1) Generalizability, applicable to various 3D semantic scene completion approaches, including 2D-3D lifting and 3D-2D transformer methods. (2) Effectiveness, as demonstrated through experiments on SemanticKITTI and NYUv2, wherein our method achieves up to 85% of the fully-supervised performance using only 10% labeled data. This approach not only reduces the cost and labor associated with data annotation but also demonstrates the potential for broader adoption in camera-based systems for 3D semantic occupancy prediction.          </p>
        </div>
      </div>
    </div>
  </div>
</section>
<!-- End paper abstract -->


<section class="section">
  <div class="container is-max-desktop">
    <!-- Method. -->
    <div class="columns is-centered has-text-centered">
      <div class="column is-four-fifths">
        <h2 class="title is-3">How it works</h2>
        <div class="content has-text-justified">
          <h3 class="title has-text-centered">
            Semi-supervised Semantic Scene Completion (Semi-SSC) Formulation
          </h3>

          <p>
            Given a sequence $\mathbf{Q}_t = \{I_{t-k}, I_{t-k+1}\ldots, I_{t-1}, I_t \}$ of $k$ consecutive frames, the SSC model $f_\theta$ generates a semantic occupancy grid which is defined in the coordinate system of the ego vehicle at the timestamp $t$. Each voxel of the grid is categorized as either empty or occupied by a specific semantic class. The grid can be obtained as follows:
            $\mathbf{\hat{Y}}_t = f_\theta(\mathbf{Q_t})$ where $\mathbf{\hat{Y}}_t \in \mathbb{R}^{H \times W \times Z \times (C+1)}$. $H$, $W$, and $Z$ denote the voxel grid's height, width,
            and depth, and $C$ is the number of the semantic classes. In a semi-supervised setting (\Approach), the dataset contains two non-overlapping subsets:
            <ul>
              <li>Labeled data $\mathcal{D}^L=\{(\mathbf{Q}_t, \mathbf{Y}_t)\}_{t=1}^L$ where each image $I_t$ has a corresponding 3D occupancy ground-truth $\mathbf{Y}_t$.</li>
              <li>Unlabeled data $\mathcal{D}^U = \{\mathbf{Q}_t\}_{t=1}^U$, which only contains images with \textbf{no LiDAR} and the amount of the unlabeled data is significantly larger than labeled data, i.e, $U \gg L$.</li>
            </ul>

          </p>

          <h3 class="title has-text-centered">
            VFG-SSC pipeline
          </h3>
          <p>
            A common strategy for tackling the semi-supervised problem is Self-Training. This involves first training a supervised model $f_\theta$ on labeled data $\mathcal{D}^L$ and then generating pseudo-labels for the unlabeled dataset $\mathcal{D}^U$. After that, the model $f_\theta$ is retrained using both the labeled and pseudo-labeled data.
            Based on this Self-Training approach, our key contribution is to enhance the quality of the pseudo-label by incorporating 3D priors extracted from 2D vision foundation models. We now outline our three-step process in detail.
          </p>
          <img id="method_train" width="100%" src="./static/images/framework.png" alt="VFG-SSC Framework"/>

          <h3 class="title has-text-centered">
            Comparison with other methods
          </h3>

          <p>
            Quantitative comparison of <span class="methodname">VFG-SSC</span> with SOTA metric depth estimators
            on several zero-shot benchmarks.
            Our VFG-SSC significantly outperforms baseline methods. Remarkably, our approach achieves comparable results to fully supervised counterparts, with just a 15% performance gap using 10% labeled data.
          </p>

          <img id="comparison" width="100%" src="./static/images/semkitti.png" alt="Comparison with other methods"/>

          <p>
            For NYUv2, we consistently outperform other methods with both 5\% and 10\% of training data, underscoring the superiority of our approach over strong baselines. These results indicate that our VFG-SSC is generalizable to different architectures, can be applied to various labeled settings, and applies to both outdoor and indoor scenarios.
          </p>
          <img id="comparison" width="100%" src="./static/images/nyu.png" alt="Comparison with other methods"/>

          <p>
            Moreover, on the SemanticKITTI hidden test set, our method compares favorably with some fully supervised methods like MonoScene, despite utilizing only 10% labeled occupancy annotation. This emphasizes the effectiveness of our 3D clues and enhancement module, which is effective in generating high-quality pseudo-labels for training any SSC backbones.
          </p>

          <img id="comparison" width="100%" src="./static/images/semkitti_test.png" alt="Comparison with other methods"/>

          <!-- <p class="subtitle mt-5">
            Refer to the pdf paper linked above for more details on qualitative, quantitative, and ablation studies.
          </p> -->
        </div>

      </div>
    </div>
    <!--/ Method. -->
  </div>
</section>


<!-- Paper poster -->
<!-- <section class="hero is-small is-light">
  <div class="hero-body">
    <div class="container">
      <h2 class="title">Poster</h2>

      <iframe  src="static/pdfs/sample.pdf" width="100%" height="550">
          </iframe>

      </div>
    </div>
  </section> -->
<!--End paper poster -->


<!--BibTex citation -->
  <section class="section" id="BibTeX">
    <div class="container is-max-desktop content">
      <h2 class="title">BibTeX</h2>
      <pre><code>@misc{pham2025semisupervised3dsemanticscene,
        title={Semi-supervised 3D Semantic Scene Completion with 2D Vision Foundation Model Guidance},
        author={Duc-Hai Pham and Duc-Dung Nguyen and Anh Pham and Tuan Ho and Phong Nguyen and Khoi Nguyen and Rang Nguyen},
        year={2025},
        eprint={2408.11559},
        archivePrefix={arXiv},
        primaryClass={cs.CV},
        url={https://arxiv.org/abs/2408.11559},
  }</code></pre>
    </div>
</section>
<!--End BibTex citation -->


  <footer class="footer">
  <div class="container">
    <div class="columns is-centered">
      <div class="column is-8">
        <div class="content">

          <p>
            This page was built using the <a href="https://github.com/eliahuhorwitz/Academic-project-page-template" target="_blank">Academic Project Page Template</a> which was adopted from the <a href="https://nerfies.github.io" target="_blank">Nerfies</a> project page.
            You are free to borrow the source code of this website, we just ask that you link back to this page in the footer. <br> This website is licensed under a <a rel="license"  href="http://creativecommons.org/licenses/by-sa/4.0/" target="_blank">Creative
            Commons Attribution-ShareAlike 4.0 International License</a>.
          </p>

        </div>
      </div>
    </div>
  </div>
</footer>

<!-- Statcounter tracking code -->

<!-- You can add a tracker to track page visits by creating an account at statcounter.com -->

    <!-- End of Statcounter Code -->

  </body>
  </html>