-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathindex.html
358 lines (307 loc) · 14.7 KB
/
index.html
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
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
<!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="KEYWORDS SHOULD BE PLACED HERE">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Academic Project Page</title>
<link rel="icon" type="image/x-icon" href="static/images/endeleze.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>
</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">Distillation Improves Visual Place Recognition for Low-Quality Queries</h1>
<div class="is-size-5 publication-authors">
<!-- Paper authors -->
<span class="author-block">
<a href="https://www.linkedin.com/in/anbang-yang/" target="_blank">Anbang Yang</a>
<span class="author-block">
<a href="https://engineering.nyu.edu/faculty/yao-wang" target="_blank">Yao Wang</a>
<span class="author-block">
<a href="https://med.nyu.edu/faculty/johnross-rizzo" target="_blank">John-Ross Rizzo</a>
</span>
<span class="author-block">
<a href="https://engineering.nyu.edu/faculty/chen-feng" target="_blank">Chen Feng</a>
</span>
</div>
<div class="is-size-5 publication-authors">
<span class="author-block">New York University,<br>Brooklyn, NY 11201, USA</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/<ARXIV PAPER ID>.pdf" 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/ai4ce/VPR4LQQ" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>Code</span>
</a>
</span>
<!-- ArXiv abstract Link -->
<span class="link-block">
<a href="https://arxiv.org/abs/<ARXIV PAPER ID>" 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/videos/VPR4LQQ.mp4"
type="video/mp4">
</video>
<h2 class="subtitle has-text-justified">
<b>VPR4LQQ:</b> We introduce a distillation neural network approach, enhancing the performance of compressed images in Visual Place Recognition. Using the Pitts250k dataset and our custom indoor dataset, we demonstrate the efficacy of our model, especially when combined with the Inter-Channel Knowledge Distillation (ICKD) loss. The video showcases a comparative animation, highlighting our model's superiority over the NetVLAD baseline in various scenarios.
</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>
The shift to online computing for real-time visual localization often requires streaming query images/videos to a server for visual place recognition (VPR), where fast video transmission may result in reduced resolution or increased quantization. This compromises the quality of global image descriptors, leading to decreased VPR performance.
To improve the low recall rate for low-quality query images, we present a simple yet effective method that uses high-quality queries only during training to distill better feature representations for deep-learning-based VPR, such as NetVLAD. Specifically, we use mean squared error (MSE) loss between the global descriptors of queries with different qualities, and inter-channel correlation knowledge distillation (ICKD) loss over their corresponding intermediate features.
We validate our approach using the both Pittsburgh 250k dataset and our own indoor dataset with varying quantization levels. By fine-tuning NetVLAD parameters with our distillation-augmented losses, we achieve notable VPR recall-rate improvements over low-quality queries, as demonstrated in our extensive experimental results. We believe this work not only pushes forward the VPR research but also provides valuable insights for applications needing dependable place recognition under resource-limited conditions.
</p>
</div>
</div>
</div>
</div>
</section>
<!-- End paper abstract -->
<!-- Image carousel -->
<section class="hero is-small">
<div class="hero-body">
<div class="container">
<div id="results-carousel" class="carousel results-carousel">
<div class="item">
<!-- Your image here -->
<img src="static/images/00356.png" alt="MY ALT TEXT"/>
</div>
<div class="item">
<!-- Your image here -->
<img src="static/images/00559.png" alt="MY ALT TEXT"/>
</div>
<div class="item">
<!-- Your image here -->
<img src="static/images/01202.png" alt="MY ALT TEXT"/>
</div>
<div class="item">
<!-- Your image here -->
<img src="static/images/01449.png" alt="MY ALT TEXT"/>
</div>
</div>
</div>
</div>
</section>
<!-- End image carousel -->
<!-- Method description -->
<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">Method</h2>
<div class="content has-text-justified">
<p>
Our approach to Visual Place Recognition (VPR) leverages knowledge distillation to make a student network mimic a sophisticated teacher network. Rooted in the NetVLAD algorithm, we employ a dual-branch distillation model comprising a student and teacher branch, both utilizing the VGG-16 architecture. The student branch processes low-quality images, while the teacher works with high-quality images. To assess the student's performance, we introduced three loss functions: Inter-channel Correlation Knowledge Distillation (ICKD) Loss, Mean Squared Error (MSE) Loss, and a Weakly Supervised Triplet Ranking Loss. However, our experiments indicated that a composite loss, excluding the triplet loss, yielded the best results. The final loss function is a weighted sum of ICKD and MSE losses.
</p>
</div>
</div>
</div>
</div>
</section>
<!-- End method description -->
<!-- Method Image -->
<section class="hero is-small">
<div class="hero-body">
<div class="columns is-centered has-text-centered">
<img src="static/images/Method.png" alt="Method Illustration">
</div>
</div>
</div>
</div>
</section>
<!-- End Method Image -->
<!-- Experiment and Result -->
<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">Experiment and Result</h2>
<div class="content has-text-justified">
<p>
We validated our distillation model using the Pitts250k dataset, a cornerstone in VPR research, and our custom dataset curated from the 6th floor of the Lighthouse Guild. The Pitts250k dataset was segmented for training, database, and validation purposes and underwent downsampling to distinct resolutions. Our dataset emphasizes the effects of video bitrate and resolution, involving extensive video processing, including downsampling and quantization. Post-processing, frames were extracted, and ground truth locations were ascertained using a combination of OpenVSLAM and the Aligner GUI.
</p>
</div>
</div>
</div>
</div>
</section>
<!-- End Experiment and Result -->
<!-- Result -->
<section class="hero is-small">
<div class="hero-body">
<img src="static/images/Figure4.JPEG" alt="Result figure">
</div>
</div>
</div>
</section>
<!-- End Result -->
<!-- Experiment and Result -->
<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">
<div class="content has-text-justified">
<p>
The following videos and figures presents heatmaps indicating the regions of interest for four models trained using different loss settings. Observably, our model (MSE+ICKD loss) predominantly concentrates on critical regions while overlooking areas prone to visual aliasing. This focused attention might underpin the superior performance exhibited by this particular loss setting.
</p>
</div>
</div>
</div>
</div>
</section>
<!-- End Experiment and Result -->
<!-- Video carousel -->
<section class="hero is-small">
<div class="hero-body">
<div class="container">
<div id="results-carousel" class="carousel results-carousel">
<div class="item item-video1">
<video poster="" id="video1" autoplay controls muted loop height="100%"> -->
<!-- Your video file here -->
<source src="static/videos/000.mp4"
type="video/mp4">
</video>
</div>
<div class="item item-video2">
<video poster="" id="video2" autoplay controls muted loop height="100%">
<!-- Your video file here -->
<source src="static/videos/120.mp4"
type="video/mp4">
</video>
</div>
<div class="item item-video3">
<video poster="" id="video3" autoplay controls muted loop height="100%">\
<!-- Your video file here -->
<source src="static/videos/240.mp4"
type="video/mp4">
</video>
</div>
</div>
</div>
</div>
</section>
<!-- End video carousel -->
<!-- Attention Map -->
<section class="hero is-small">
<div class="hero-body">
<img src="static/images/Figure6.png" alt="Attention Map">
</div>
</div>
</div>
</section>
<!-- End Attention Map -->
<!-- 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>BibTex Code Here</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.
<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>