-
Notifications
You must be signed in to change notification settings - Fork 0
/
index.html
278 lines (245 loc) · 13.4 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
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta name="description"
content="Zero-shot Model Diagnosis that generates counterfactual samples and sensitivity analysis.">
<meta name="keywords" content="Trustworthy Machine Learning">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Zero-shot Model Diagnosis</title>
<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>
<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"><span style="color: #990000;">Z</span>er<span style="color: #990000;">o</span>-sh<span style="color: #990000;">o</span>t <span style="color: #990000;">M</span>odel Diagnosis (ZOOM)</h1>
<h3 class="title is-4 publication-title">CVPR 2023</h3>
<div class="is-size-5 publication-authors">
<span class="author-block">
<a href="https://peterljq.github.io/">Jinqi Luo</a><sup>*</sup>,</span>
<span class="author-block">
<a href="https://www.zhaoningwang.com">Zhaoning Wang</a><sup>*</sup>,</span>
<span class="author-block">
<a href="https://github.com/ChenWu98">Chen Henry Wu</a>,
</span>
<span class="author-block">
<a href="https://www.donghuang-research.com/">Dong Huang</a>,
</span>
<span class="author-block">
<a href="https://www.cs.cmu.edu/~ftorre/">Fernando De La Torre</a>
</span>
</div>
<div class="is-size-5 publication-authors">
<span class="author-block">School of Computer Science<br/>Carnegie Mellon University</span>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<!-- PDF Link. -->
<span class="link-block">
<a href="https://zero-shot-model-diagnosis.github.io/ZOOM.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>
<span class="link-block">
<a href="https://arxiv.org/abs/2303.15441"
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>
<!-- Code Link. -->
<span class="link-block">
<a href="https://github.com/humansensinglab/ZOOM"
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>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="hero teaser">
<div class="container is-max-desktop" style="text-align:center">
<h2 class="subtitle has-text-centered">
How can we diagnose a deep learning computer vision model <b>without a test set</b>?
</h2>
<img src="./static/images/teaser_figure.png" style="width:60%;">
<h2 class="subtitle has-text-centered">
Diagnose your vision model's failure by just typing the attributes of interest. <br/>Our Plug-and-play framework generates a histogram of sensitivity analysis.
</h2>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<!-- Abstract. -->
<div class="columns is-centered">
<div class="column is-full-width">
<h2 class="title is-3">Abstract</h2>
<div class="content has-text-justified">
<p>
When it comes to deploying deep vision models, the behavior of these systems must be explicable to ensure confidence in their reliability and fairness. A common approach
to evaluate deep learning models is to build a labeled test
set with attributes of interest and assess how well it performs. However, creating a balanced test set (i.e., one that
is uniformly sampled over all the important traits) is often time-consuming, expensive, and prone to mistakes. The
question we try to address is: can we evaluate the sensitivity of deep learning models to arbitrary visual attributes
<b>without an annotated test set?</b>
</p>
<p>
This paper argues the case that <b><span style="color: #990000;">Z</span>er<span style="color: #990000;">o</span>-sh<span style="color: #990000;">o</span>t <span style="color: #990000;">M</span>odel Diagnosis</b> (ZOOM) is possible without the need for a test set nor
labeling. To avoid the need for test sets, our system relies
on a generative model and CLIP. The key idea is enabling
the user to select a set of prompts (relevant to the problem) and our system will automatically search for semantic counterfactual images (i.e., synthesized images that flip
the prediction in the case of a binary classifier) using the
generative model. We evaluate several visual tasks (classification, key-point detection, and segmentation) in multiple
visual domains to demonstrate the viability of our methodology. Extensive experiments demonstrate that our method
is capable of producing counterfactual images and offering
sensitivity analysis for model diagnosis without the need for
a test set.
</p>
</div>
<br/>
<h2 class="title is-3">Framework</h2>
<div class="columns is-vcentered interpolation-panel">
<img src="./static/images/overall_pipeline.jpg" style="width:100%;">
</div>
<div class="content has-text-justified">
<p>
<b>The ZOOM framework.</b> Black solid lines stand for forward passes, red dashed lines stand for backpropagation, and purple
dashed lines stands for inference after the optimization converges. The user inputs single or multiple attributes, and we map them into
edit directions. Then we assign to each edit direction (attribute) a weight, which represents how much we are adding/removing this attribute. We iteratively perform adversarial learning on the attribute space to maximize the counterfactual
effectiveness. More details in the <a href="https://github.com/" target="_blank">paper</a>.
</p>
</div>
<br/>
<!--/ Interpolating. -->
<h2 class="title is-3">Diagnosis Visualization</h2>
<h3 class="title is-4">Counterfactual demo</h3>
<div class="content has-text-justified">
<p>
This demo shows how a classifier score changes with semantic modifications of the image. We refer the images as counterfactual images, as the images are generated to flip the target model, in this case, a classifier.
<!-- Some progressive generation of counterfactual samples (images which flip the prediction of the target model). -->
</p>
</div>
<table>
<tr>
<td><img src="./static/images/gif/animation.gif"></td>
</tr>
</table>
<div class="content has-text-justified">
<p>
The first four images are on Dog/Cat classifier, and the following two are on perceived age (Young) classifier. Lastly, the final two are on the perceived gender classifier.
</p>
</div>
<br/>
<h3 class="title is-4">Single-attribute search</h3>
<div class="content has-text-justified">
<p>
Model diagnosis histograms generated by the counterfactual's effectiveness on the target model for a particular attribute. The vertical axis values reflect the attribute sensitivities calculated by averaging the model probability change over all sampled images. The horizontal axis is the attribute space input by user.
<!-- Single-attribute counterfactual to generate sensitivity histograms. Model diagnosis histograms generated by the counterfactual's effectiveness on the target model. The vertical axis values reflect the attribute sensitivities calculated by averaging the model probability change over all sampled images. The horizontal axis is the attribute space input by user. -->
</p>
</div>
<div style="text-align:center">
<img class="summary-img" src="./static/images/histogram_with_image.png" style="width:100%;"> <br>
</div>
<br/>
<h3 class="title is-4">Multi-attribute search</h3>
<div class="content has-text-justified">
<p>
This demo shows the generation of counterfactual images over multiple simultaneous attributes.
In the images with a box on the right-up corner, the number indicates the probability of the classifier.
<!-- Multi-attribute counterfactual search in the vehicle, church, and human face domain. In this case, Multiple attributes are edited and combined to generate the most powerful counterfacutal samples. If exists, the right up corner box in each image records the model predication (e.g., probability). -->
</p>
</div>
<!-- <div class="content has-text-justified">
<p>
Counterfactual images on binary classifiers, the number indicates the probability of the classifier.
</p>
</div> -->
<div style="text-align:center">
<h4 class="title is-6">Human Computer Vision models:</h4>
<img class="summary-img" src="./static/images/multi_attr_human.png" style="width:100%;"> <br>
<br>
</div>
<p>
We also show additional counterfactuals examples on other computer vision models, including semantic segmentation, multi-class classification, and binary church classifier (BCC).
<!-- Multi-attribute counterfactual search in the vehicle, church, and human face domain. In this case, Multiple attributes are edited and combined to generate the most powerful counterfacutal samples. If exists, the right up corner box in each image records the model predication (e.g., probability). -->
</p>
<div style="text-align:center">
<h5 class="title is-6">Other Computer Vision models:</h5>
<img class="summary-img" src="./static/images/multi_attr_additional.png" style="width:72%;"> <br>
</div>
</div>
</div>
</div>
</section>
<section class="section" id="BibTeX">
<div class="container is-max-desktop content">
<h2 class="title is-3">BibTeX</h2>
<pre><code>@inproceedings{luo2023zeroshot,
title={Zero-shot Model Diagnosis},
author={Jinqi Luo and Zhaoning Wang and Chen Henry Wu and Dong Huang and Fernando De La Torre},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
year={2023}
}
</code></pre>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<!-- Abstract. -->
<div class="columns is-centered">
<div class="column is-full-width">
<h2 class="title is-3">Acknowledgements</h2>
<div class="content has-text-justified">
<p>We would like to thank George Cazenavette, Tianyuan Zhang, Yinong Wang, Hanzhe Hu, Bharath Raj for their invaluable feedbacks and suggestions in paper presentation and experimental design.
We sincerely express our gratitude to Ken Ziyu Liu, Jiashun Wang, Bowen Li, and Ce Zheng for their revise and support that improve this work.</p>
</div>
</div>
</div>
</div>
</section>
<footer class="footer">
<div class="columns is-centered">
<div class="column is-8">
<div class="content has-text-justified">
<p>
We thank <a href="https://github.com/nerfies/nerfies.github.io">Nerfies</a> for opensouring the template of this website.
The website theme colors are derived from [<a href="https://arxiv.org/pdf/2111.06377.pdf">1</a>, <a href="https://branding.web-resources.upenn.edu/logos-and-branding/elements-penn-logo">2</a>].
</p>
</div>
</div>
</div>
</footer>
<a href="https://clustrmaps.com/site/1btnr" title="Visit tracker"><img src="//www.clustrmaps.com/map_v2.png?d=W7qxge2Snq-QqfwiBAVe4ve5n7nnPGW2h2BgMuI1Bsk&cl=ffffff" hidden/></a>
</body>
</html>