-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathindex.html
335 lines (291 loc) · 13.2 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
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta name="description"
content="BigSmall">
<meta name="keywords" content="BigSmall, Physiological Sensing, Facial Action Units, Photoplethysmography, Respiration">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>
BigSmall
</title>
<!-- 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="./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">
<link rel="icon" href="./static/images/favicon.svg">
<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>
</head>
<body>
<!-- <nav class="navbar" role="navigation" aria-label="main navigation">
<div class="navbar-brand">
<a role="button" class="navbar-burger" aria-label="menu" aria-expanded="false">
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
</a>
</div>
</div>
</div>
</nav> -->
<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">BigSmall:</h1>
<h3 class="title is-3 publication-title">Efficient Multi-Task Learning <br> For Physiological Measurements</h3>
<div class="is-size-5 publication-authors">
<span class="author-block">
<a href="https://girishvn.github.io/">Girish Narayanswamy</a><sup>1</sup>,</span>
<span class="author-block">
<a href="https://www.linkedin.com/in/yujianancyliu/">Yujia (Nancy) Liu</a><sup>1</sup>,</span>
<span class="author-block">
<a href="https://www.mit.edu/~yuzhe/">Yuzhe Yang</a><sup>2</sup>,
</span>
<span class="author-block">
<a href="https://www.linkedin.com/in/chengqian-ma-356566162/">Chengqian (Jack) Ma</a><sup>1</sup>,
</span>
<br>
<span class="author-block">
<a href="https://xliucs.github.io/">Xin Liu</a><sup>1</sup>,
</span>
<span class="author-block">
<a href="http://alumni.media.mit.edu/~djmcduff/"> Daniel McDuff</a><sup>1</sup>,
</span>
<span class="author-block">
<a href="https://www.cs.washington.edu/people/faculty/shwetak">Shwetak Patel</a><sup>1</sup>
</span>
</div>
<div class="is-size-5 publication-authors">
<span class="author-block"><sup>1</sup>University of Washington,</span>
<span class="author-block"><sup>2</sup>Massachusetts Institute of Technology</span>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<!-- PDF Link. -->
<span class="link-block">
<a href="./static/paper/BigSmall_Arxiv.pdf"
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="https://github.com/girishvn/BigSmall"
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 Link. -->
<span class="link-block">
<a href="https://arxiv.org/abs/2303.11573"
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>
<section class="hero teaser">
<div class="container is-max-desktop">
<div class="hero-body">
<h3 class="subtitle has-text-centered">
<span class="dnerf">BigSmall</span> is an efficient network architecture for multi-task physiological sensing. The network concurrently derives
photoplethysmography (pulse), respiration, and action unit signals.
</h3>
<div class="teaser-vids">
<video id="teaser1" autoplay muted loop playsinline height="100%">
<source src="./static/videos/bigsmall_ex1.mp4" type="video/mp4">
</video>
<video id="teaser2" autoplay muted loop playsinline height="100%">
<source src="./static/videos/bigsmall_ex2.mp4" type="video/mp4">
</video>
</div>
<h3>
AU06 (Cheek Raiser), AU07 (Lid Tightener), AU10 (Upper Lip Raiser),
AU12 (Lip Corner Puller), AU14 (Dimpler)
</h3>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<!-- Abstract. -->
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h3 class="title is-3">Abstract</h3>
<div class="content has-text-justified">
<p>
Understanding of human visual perception has historically inspired the design of computer vision
architectures. As an example, perception occurs at different scales both spatially and temporally,
suggesting that the extraction of salient visual information may be made more effective by paying
attention to specific features at varying scales. Visual changes in the body due to physiological
processes also occur at different scales and with modality-specific characteristic properties.
</p>
<p>
Inspired by this, we present BigSmall, an efficient architecture for physiological and behavioral measurement.
We present the first joint camera-based facial action, cardiac, and pulmonary measurement model.
We propose a multi-branch network with wrapping temporal shift modules that yields both accuracy
and efficiency gains. We observe that fusing low-level features leads to suboptimal performance,
but that fusing high level features enables efficiency gains with negligible loss in accuracy.
Experimental results demonstrate that BigSmall significantly reduces the computational costs.
Furthermore, compared to existing task-specific models, BigSmall achieves comparable or better
results on multiple physiological measurement tasks simultaneously with a unified model.
</p>
</div>
</div>
</div>
<!--/ Abstract. -->
<!--/ Paper video. -->
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<!-- Model Arch. -->
<div class="columns is-centered">
<div class="column is-full-width">
<h3 class="title is-3">Model Architecture</h3>
<div class="content has-text-justified">
<p>
<span class="dnerf">BigSmall</span> leverages a dual branch architecture, consisting of a Big branch to model
high fidelity spatial features using high-resolution raw-frame inputs, and a Small branch to model temporal dynamics
using low-resolution difference-frame inputs.
</p>
<img src="./static/images/BS_Network_Arch.svg" alt="BS">
</div>
</div>
</div>
<!-- Computational Efficiency -->
<div class="columns is-centered">
<div class="column is-full-width">
<h3 class="title is-3">Computational Efficiency: Big Branch Downsampling</h3>
<div class="content has-text-justified">
<p>
<span class="dnerf">BigSmall</span> frames the Big branch as a means to model relatively low-frequency
spatial signals, while the Small branch models relatively high-frequency temporals.
<span class="dnerf">BigSmall</span> is thus able to temporally-downsample the Big branch inputs to a fraction of the
inputs seen by the Small branch. As compute of the model is driven by the Big branch convolutions, this downsampling
results in a significant compute benefit. Sepcifically, if the Big branch sees 1/N frames seen by the Small branch
the compute is reduced by a factor of approximately N.
</p>
<img src="./static/images/BigBranch_DownSample.svg" alt="BigDS">
</div>
</div>
</div>
<!-- Temporal Modeling -->
<div class="columns is-centered">
<div class="column is-full-width">
<h3 class="title is-3"> Efficiencient Temporal Modeling: Small Branch WTSMs</h3>
<div class="content has-text-justified">
<p>
<span class="dnerf">BigSmall</span> introduces Wrapping Temporal Shift Modules (WTSM), which allow for inter-frame
information sharing, thus allowing for more robhust temporal modeling in the Small branch. Unlike traditional
<a href="https://arxiv.org/abs/1811.08383">TSMs</a>, WTSMs wrap shifted-out frame channels to eliminate zero-filled
feature maps. This allows for augmented temporals even when latency constraints are high or training-regiments
neccessitate high training batch variance.
</p>
<img src="./static/images/WTSM.svg" alt="WTSM">
</div>
</div>
</div>
<!-- Concurrent Work. -->
<div class="columns is-centered">
<div class="column is-full-width">
<h3 class="title is-3">Related Links</h3>
<div class="content has-text-justified">
<p>
The data pre-processsing scripts, machine-learning pipeline, network architevtures, evaluation methods, and
pre-trained models can be found at our <a href="https://github.com/ubicomplab/rPPG-Toolbox">GitHub Repository</a>.
This code is licensed under the MIT License. If you end up making use of any of these material please make sure to cite our paper!
</p>
<p>
The machine learning pipeline for this work as been adapted from the
<a href="https://github.com/ubicomplab/rPPG-Toolbox">rPPG-Toolbox</a>, a project to make deep-physiological sensing
research more standardized and accessible. The publication for this work can be found
<a href="https://arxiv.org/abs/2210.00716">here</a>.
</p>
</div>
</div>
</div>
<!--/ Concurrent Work. -->
</div>
</section>
<section class="section" id="BibTeX">
<div class="container is-max-desktop content">
<h3 class="title is-3">Citation</h3>
<pre><code>
@misc{narayanswamy2023bigsmall,
title={BigSmall: Efficient Multi-Task Learning for Disparate Spatial and Temporal Physiological Measurements},
author={Girish Narayanswamy and Yujia Liu and Yuzhe Yang and Chengqian Ma and Xin Liu and Daniel McDuff and Shwetak Patel},
year={2023},
eprint={2303.11573},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
</code></pre>
</div>
</section>
<footer class="footer">
<div class="container">
<div class="content has-text-centered">
<a class="icon-link"
href="./static/paper/BigSmall_Arxiv.pdf">
<i class="fas fa-file-pdf"></i>
</a>
<a class="icon-link" href="https://github.com/girishvn/BigSmall" class="external-link" disabled>
<i class="fab fa-github"></i>
</a>
</div>
<div class="columns is-centered">
<div class="column is-8">
<div class="content">
<p>
This website is licensed under a
<a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">
Creative Commons Attribution-ShareAlike 4.0 International License</a>.
</p>
<p>
The source code for this site was adapted from
<a href="https://github.com/girishvn/BigSmall">this work</a>.
If you decide to use this web template please make sure to link their work!
</p>
</div>
</div>
</div>
</div>
</footer>
</body>
</html>