-
Notifications
You must be signed in to change notification settings - Fork 0
/
index.html
248 lines (237 loc) · 10.1 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
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8" />
<title>
Davidsonian Scene Graph: Improving Reliability in Fine-Grained Evaluation for Text-to-Image
Generation
</title>
<meta name="description" content="DSG (2023)" />
<link rel="preconnect" href="https://fonts.googleapis.com" />
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />
<link
href="https://fonts.googleapis.com/css2?family=Open+Sans:wght@300;400&display=swap"
rel="stylesheet"
/>
<link
href="https://fonts.googleapis.com/css?family=Material+Icons&display=block"
rel="stylesheet"
/>
<link rel="stylesheet" href="style.css" />
<link
rel="stylesheet"
href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/4.7.0/css/font-awesome.min.css"
/>
<!-- Global site tag (gtag.js) - Google Analytics -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-H3R11ZELBV"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag() {
dataLayer.push(arguments);
}
gtag("js", new Date());
gtag("config", "G-H3R11ZELBV");
</script>
</head>
<body onload="loadCallback();">
<div class="mask" id="start-mask"></div>
<div class="wrap">
<h1>
Davidsonian Scene Graph: Improving Reliability in Fine-Grained Evaluation for Text-to-Image
Generation
</h1>
<div class="author-grid-top">
<div class="author-column">
<span class="author"><a href="https://j-min.io/" target="_blank">Jaemin Cho</a></span
><br />
<span class="affiliation">UNC Chapel Hill</span>
</div>
<div class="author-column">
<span class="author"
><a href="https://yushi-hu.github.io/" target="_blank">Yushi Hu</a></span
><br />
<span class="affiliation">UWashington</span>
</div>
<div class="author-column">
<span class="author"
><a href="https://www.roopalgarg.com/" target="_blank">Roopal Garg</a></span
><br />
<span class="affiliation">Google Research</span>
</div>
<div class="author-column">
<span class="author"
><a href="https://panderson.me/" target="_blank">Peter Anderson</a></span
><br />
<span class="affiliation">Google Research</span>
</div>
<div class="author-column">
<span class="author"
><a href="https://ranjaykrishna.com/index.html" target="_blank">Ranjay Krishna</a></span
><br />
<span class="affiliation">UWashington</span>
</div>
</div>
<div class="author-grid-bottom">
<div class="author-column">
<span class="author"
><a href="https://research.google/people/JasonBaldridge/" target="_blank"
>Jason Baldridge</a
></span
><br />
<span class="affiliation">Google Research</span>
</div>
<div class="author-column">
<span class="author"
><a href="https://www.cs.unc.edu/~mbansal/" target="_blank">Mohit Bansal</a></span
><br />
<span class="affiliation">UNC Chapel Hill</span>
</div>
<div class="author-column">
<span class="author"
><a href="https://jponttuset.cat" target="_blank">Jordi Pont-Tuset</a></span
><br />
<span class="affiliation">Google Research</span>
</div>
<div class="author-column">
<span class="author"
><a href="https://research.google/people/107321/" target="_blank">Su Wang</a></span
><br />
<span class="affiliation">Google Research</span>
</div>
</div>
<div class="button-bar center">
<a href="https://arxiv.org/abs/2310.18235" target="_blank"
><span class="button"
><span class="button-wrapper"
><i style="font-size: small" class="fa fa-file"></i> PDF</span
></span
></a
>
<a href="https://github.com/j-min/DSG" target="_blank"
><span class="button"
><span class="button-wrapper"><i class="fa fa-github"></i> Code</span></span
></a
>
<a href="https://github.com/j-min/DSG/blob/main/dsg-1k-anns.csv" target="_blank">
<span class="button"
><span class="button-wrapper"
><i style="font-size: small" class="fa fa-database"></i> DSG-1k</span
></span
>
</a>
</div>
<div>
<img class="splash-figure" src="./web-data/figures/dsg_splash.jpg" />
</div>
<hr class="separator" />
<h2>Abstract</h2>
<div class="block">
Evaluating text-to-image models is notoriously difficult. A strong recent approach for
assessing text-image faithfulness is based on QG/A (question generation and answering),
which uses pre-trained foundational models to automatically generate a set of questions and
answers from the prompt, and output images are scored based on whether these answers
extracted with a visual question answering model are consistent with the prompt-based
answers. This kind of evaluation is naturally dependent on the quality of the underlying QG
and QA models. We identify and address several reliability challenges in existing QG/A work:
(a) QG questions should respect the prompt (avoiding hallucinations, duplications, and
omissions) and (b) VQA answers should be consistent (not asserting that there is no
motorcycle in an image while also claiming the motorcycle is blue). We address these issues
with Davidsonian Scene Graph (DSG), an empirically grounded evaluation framework inspired by
formal semantics. DSG is an automatic, graph-based QG/A that is modularly implemented to be
adaptable to any QG/A module. DSG produces atomic and unique questions organized in
dependency graphs, which (i) ensure appropriate semantic coverage and (ii) sidestep
inconsistent answers. With extensive experimentation and human evaluation on a range of
model configurations (LLM, VQA, and T2I), we empirically demonstrate that DSG addresses the
challenges noted above. Finally, we present DSG-1k, an open-sourced evaluation benchmark
that includes 1,060 prompts, covering a wide range of fine-grained semantic categories with
a balanced distribution. We will release the DSG-1k prompts and the corresponding DSG
questions.
</div>
<hr class="separator" />
<h2>QG/A: New Paradigm in T2I Alignment Eval</h2>
<div>
<img class="figure-row" src="./web-data/figures/t2i_eval_paradigm_shift.jpg" />
</div>
<hr class="separator" />
<h2>Reliability Issues in Existing QG/A Methods</h2>
<div class="center">
<img style="max-width: 70%" src="./web-data/figures/t2i_eval_reliability_issues.jpg" />
</div>
<hr class="separator" />
<h2>DSG Solution to the Reliability Issues</h2>
<div>
<img class="figure-row" src="./web-data/figures/t2i_eval_dsg_solution.jpg" />
</div>
<hr class="separator" />
<h2>Publication</h2>
<div class="center">
<b
>Davidsonian Scene Graph: Improving Reliability in Fine-Grained Evaluation for
Text-to-Image Generation</b
><br />
Jaemin Cho, Yushi Hu, Roopal Garg, Peter Anderson, Ranjay Krishna,<br />Jason Baldridge,
Mohit Bansal, Jordi Pont-Tuset, Su Wang<br />
ICLR 2024<br />
[<a href="https://arxiv.org/abs/2310.18235" target="_blank">PDF</a>] [<a
href="javascript:document.getElementById('bibtex').open = true;"
>BibTeX</a
>]
<a href="https://arxiv.org/abs/2310.18235" target="_blank">
<div class="row paper-row">
<div class="paper-col">
<img class="paper-col-im" src="./web-data/thumb/thumbnail-0.jpg" />
</div>
<div class="paper-col">
<img class="paper-col-im" src="./web-data/thumb/thumbnail-1.jpg" />
</div>
<div class="paper-col">
<img class="paper-col-im" src="./web-data/thumb/thumbnail-2.jpg" />
</div>
<div class="paper-col">
<img class="paper-col-im" src="./web-data/thumb/thumbnail-3.jpg" />
</div>
<div class="paper-col">
<img class="paper-col-im" src="./web-data/thumb/thumbnail-4.jpg" />
</div>
<div class="paper-col">
<img class="paper-col-im" src="./web-data/thumb/thumbnail-5.jpg" />
</div>
<div class="paper-col">
<img class="paper-col-im" src="./web-data/thumb/thumbnail-6.jpg" />
</div>
<div class="paper-col">
<img class="paper-col-im" src="./web-data/thumb/thumbnail-7.jpg" />
</div>
<div class="paper-col">
<img class="paper-col-im" src="./web-data/thumb/thumbnail-8.jpg" />
</div>
<div class="paper-col">
<img class="paper-col-im" src="./web-data/thumb/thumbnail10.jpg" />
</div>
</div>
</a>
</div>
</div>
<mwc-dialog id="bibtex">
<pre id="bibtex_text">
@inproceedings{JaeminCho2024,
author = {Jaemin Cho and Yushi Hu and Roopal Garg and Peter Anderson and Ranjay Krishna and Jason Baldridge and Mohit Bansal and Jordi Pont-Tuset and Su Wang},
title = {{Davidsonian Scene Graph: Improving Reliability in Fine-Grained Evaluation for Text-to-Image Generation}},
booktitle = {ICLR},
year = {2024}
}</pre
>
<mwc-button
slot="secondaryAction"
onclick="navigator.clipboard.writeText(document.getElementById('bibtex_text').innerText);"
>
Copy to clipboard
</mwc-button>
<mwc-button slot="primaryAction" onclick="document.getElementById('bibtex').open = false;">
Close
</mwc-button>
</mwc-dialog>
<br /><br /><br /><br /><br />
<script type="text/javascript" src="web_code.js"></script>
</body>
</html>