-
Notifications
You must be signed in to change notification settings - Fork 2
Expand file tree
/
Copy pathindex.html
More file actions
570 lines (506 loc) · 18.4 KB
/
index.html
File metadata and controls
570 lines (506 loc) · 18.4 KB
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
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="description"
content="Discover the FakeReasoning, an accurate and interpretable AI-generated images detector.">
<meta property="og:title" content="FakeReasoning: Towards Generalizable Forgery Detection and Reasoning">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>FakeReasoning</title>
<!-- Fonts and Libraries -->
<link href="https://fonts.googleapis.com/css2?family=Roboto:wght@400;500;700&display=swap" rel="stylesheet">
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/bootstrap@5.3.0/dist/css/bootstrap.min.css">
<link rel="stylesheet" href="https://unpkg.com/aos@2.3.1/dist/aos.css">
<!-- Custom Styles -->
<style>
body {
font-family: 'Roboto', sans-serif;
background: linear-gradient(to bottom, #ffffff, #ffffff);
color: #333;
}
.hero-header {
background: linear-gradient(135deg, #ffffff, #ffffff);
color: #000000;
padding: 3rem 1rem;
text-align: center;
border-radius: 1rem;
box-shadow: 0 4px 15px rgba(0, 0, 0, 0.2);
}
.hero-header h1 {
font-size: 2.2rem;
font-weight: 700;
}
h2 {
font-weight: 700;
}
.image-grid {
display: grid;
grid-template-columns: repeat(auto-fit, minmax(250px, 1fr));
gap: 1rem;
margin-top: 2rem;
}
.image-container {
position: relative;
overflow: hidden;
cursor: pointer;
border-radius: 15px;
transition: transform 0.3s ease, box-shadow 0.3s ease;
}
.image-container:hover {
transform: scale(1.05);
box-shadow: 0 10px 25px rgba(0, 0, 0, 0.2);
}
.image-container img {
width: 100%;
height: 100%;
object-fit: cover;
transition: opacity 0.5s ease;
}
.image-container .alternate {
position: absolute;
top: 0;
left: 0;
opacity: 0;
width: 100%;
height: 100%;
object-fit: cover;
}
.image-container:active .alternate {
opacity: 1;
}
.image-container:active .original {
opacity: 0;
}
.citation-box {
background-color: #ffffff;
border-left: 4px solid #ffffff;
overflow-x: auto;
}
.custom-container {
max-width: 100%;
margin: auto;
}
.custom-container img {
width: auto;
height: 350px;
object-fit: contain;
}
.carousel-control-prev-icon,
.carousel-control-next-icon {
background-image: none;
width: 20px;
height: 20px;
border: solid black;
border-width: 0 3px 3px 0;
display: inline-block;
padding: 5px;
}
.carousel-control-prev-icon {
transform: rotate(135deg);
}
.carousel-control-next-icon {
transform: rotate(-45deg);
}
.carousel-control-prev,
.carousel-control-next {
background-color: transparent;
opacity: 0.8;
transition: opacity 0.3s;
}
.carousel-control-prev:hover,
.carousel-control-next:hover {
opacity: 1;
}
.custom-width {
max-width: 60%;
margin: 0 auto;
}
.bullet-point {
position: relative;
padding-left: 20px;
margin-bottom: 10px;
}
.bullet-point::before {
content: "•";
position: absolute;
left: 0;
top: 0;
font-size: 1.5em;
line-height: 1;
}
.tab-container {
max-width: 60%;
margin: 2rem auto;
--primary-color: #2c3e50;
--active-color: #66829A;
--hover-color: #f0f4f8;
}
.tab-nav {
display: flex;
gap: 1rem;
border-bottom: 1px solid #eee;
justify-content: center;
padding-bottom: 0.5rem;
}
.tab-button {
background: none;
border: none;
padding: 12px 24px;
cursor: pointer;
font-size: 1.1rem;
color: var(--primary-color);
transition: all 0.3s ease;
border-radius: 6px 6px 0 0;
position: relative;
}
.tab-button:hover {
background: var(--hover-color);
}
.tab-button.active {
color: var(--active-color);
}
.tab-button.active::after {
content: '';
position: absolute;
bottom: -0.5rem;
left: 0;
width: 100%;
height: 3px;
background: var(--active-color);
animation: underline 0.3s ease;
}
.tab-content {
display: none;
padding: 2rem;
background: white;
box-shadow: 0 4px 6px rgba(0, 0, 0, 0.05);
border-radius: 0 8px 8px 8px;
animation: fadeIn 0.3s ease;
width: 100%;
margin: 0 auto;
}
.tab-content.active {
display: block;
}
.responsive-image {
max-width: 100%;
height: 450px;
display: block;
margin: 0 auto;
}
@keyframes underline {
from {
width: 0
}
to {
width: 100%
}
}
@keyframes fadeIn {
from {
opacity: 0;
transform: translateY(10px)
}
to {
opacity: 1;
transform: translateY(0)
}
}
@media (max-width: 480px) {
.tab-nav {
flex-wrap: wrap;
gap: 0.5rem;
}
.tab-button {
font-size: 1rem;
padding: 10px 16px;
}
}
</style>
</head>
<body>
<!-- Header Section -->
<section class="hero-header">
<div class="container">
<h1>Towards Generalizable Forgery Detection and Reasoning</h1>
<p data-aos-delay="200">
Achieving accurate and interpretable AI-generated images detection via MLLMs' visual reasoning capability.
</p>
<!-- Buttons for Code, Dataset, Demo -->
<div class="d-flex justify-content-center mt-4" data-aos-delay="300">
<!-- Arxiv Button -->
<a href="https://arxiv.org/abs/2503.21210" target="_blank" class="btn btn-dark btn-lg mx-2">
<span class="icon">
<img src="static/images/arxiv-logo-one-color-white.svg" width="25" alt="ArXiv">
</span>
Paper
</a>
<!-- Code Button -->
<a href="https://github.com/PRIS-CV/FakeReasoning" target="_blank" class="btn btn-dark btn-lg mx-2" disabled>
<span class="icon">
<img src="static/images/gitt.png" width="25" alt="GitHub">
</span>
Code
</a>
<!-- Model Button -->
<a href="https://huggingface.co/AnnaGao/FakeReasoning" target="_blank" class="btn btn-dark btn-lg mx-2" disabled>
<span class="icon">
<img src="static/images/hf-logo-pirate.png" width="25" alt="Model">
</span>
Model
</a>
<!-- Dataset Button -->
<a href="https://huggingface.co/datasets/AnnaGao/MMFR-Dataset" target="_blank" class="btn btn-dark btn-lg mx-2" disabled>
<span class="icon">
<img src="static/images/hf-logo-pirate.png" width="25" alt="Dataset">
</span>
Dataset
</a>
<!-- Demo Button (coming soon) -->
<!-- <a href="#" target="_blank" class="btn btn-dark btn-lg mx-2" disabled>
<span class="icon">
<img src="static/images/output.png" width="25" alt="Demo">
</span>
Demo (coming soon)
</a> -->
</div>
</div>
</section>
<!-- Abstract and Authors Section -->
<section class="py-5 text-center ">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h2 class="subtitle is-size-5">Authors</h2>
<div class="is-size-5 publication-authors">
<span class="author-block">Yueying Gao<sup>1</sup>,</span>
<span class="author-block">Dongliang Chang<sup>1</sup>,</span>
<span class="author-block">Bingyao Yu<sup>2</sup>,</span>
<span class="author-block">Haotian Qin<sup>1</sup>,</span>
<span class="author-block">Muxi Diao<sup>1</sup>,</span>
<span class="author-block">Lei Chen<sup>2</sup>,</span>
<span class="author-block">Kongming Liang<sup>1</sup>,</span>
<span class="author-block">Zhanyu Ma<sup>1</sup></span>
</div>
</div>
<div class="is-size-5 publication-authors">
<span class="author-block"><sup>1</sup> PRIS, Beijing University of Posts and Telecommunications</span><br>
<span class="author-block"><sup>2</sup> Tsinghua University</span><br>
</div>
</div>
</div>
</section>
<section class="py-5 ">
<div class="container custom-width">
<div class="columns is-centered">
<div class="column is-four-fifths">
<h2 class="title is-3" style="text-align: center;">Contributions</h2>
<div class="content has-text-left">
<div class="bullet-point">
<strong>FakeReasoning:</strong> A forgery detection and reasoning framework, providing accurate detection
with structured and reliable reasoning over forgery attributes.
</div>
<div class="bullet-point">
<strong>MMFR-Dataset:</strong> A multi-modal forgery reasoning dataset, containing 120K training images
and 20K evaluation images annotated with detailed reasoning over forgery attributes.
</div>
</div>
</div>
</div>
</div>
</section>
<!-- Visualization Section -->
<section class="py-5">
<div class="tab-container">
<h2 class="mb-4 text-center">Detection and Reasoning Performance of FakeReasoning</h2>
<p class="mb-4">
<strong>FakeReasoning</strong> conducts forgery reasoning task
in structured stages (summary, caption, reasoning and conclusion) and hierarchical steps (low-level and high-level),
leading to accurate and interpretable detection.
</p>
<div class="tab-nav">
<button class="tab-button active" onclick="switchTab(event, 'tab1')">GAN-generated image</button>
<button class="tab-button" onclick="switchTab(event, 'tab2')">DM-generated image</button>
<button class="tab-button" onclick="switchTab(event, 'tab3')">Real image I</button>
<button class="tab-button" onclick="switchTab(event, 'tab4')">Real image II</button>
</div>
<div id="tab1" class="tab-content active">
<img src="static/images/visualization_1.png" alt="GAN-generated image" style="max-width: 100%; height: auto;">
</object>
</div>
<div id="tab2" class="tab-content">
<img src="static/images/visualization_2.png" alt="DM-generated image" style="max-width: 100%; height: auto;">
</div>
<div id="tab3" class="tab-content">
<img src="static/images/visualization_3.png" alt="Real image I" style="max-width: 100%; height: auto;">
</div>
<div id="tab4" class="tab-content">
<img src="static/images/visualization_4.png" alt="Real image II" style="max-width: 100%; height: auto;">
</div>
</div>
</section>
<!-- Training sets -->
<section class="py-5">
<div class="custom-container text-center">
<h2 class="mb-4" >MMFR-Dataset</h2>
<p class="mb-4">
Constructed with advanced GPT-4o, <strong>MMFR-Dataset</strong> contains over 100,000 images with over
310,000 reasoning annotations as its training set. <br>
Here we illustrate several images and according annotations from reasoning stage.
</p>
<div id="mmfrCarousel" class="carousel slide" data-bs-ride="false">
<div class="carousel-inner">
<div class="carousel-item active">
<img loading="lazy" src="static/images/train_1.jpg" class="responsive-image" alt="Image 1">
</div>
<div class="carousel-item">
<img loading="lazy" src="static/images/train_2.jpg" class="responsive-image" alt="Image 2">
</div>
</div>
<button class="carousel-control-prev" type="button" data-bs-target="#mmfrCarousel" data-bs-slide="prev">
<span class="carousel-control-prev-icon" aria-hidden="true"></span>
<span class="visually-hidden">Previous</span>
</button>
<button class="carousel-control-next" type="button" data-bs-target="#mmfrCarousel" data-bs-slide="next">
<span class="carousel-control-next-icon" aria-hidden="true"></span>
<span class="visually-hidden">Next</span>
</button>
</div>
</div>
</section>
<!-- Evaluation sets -->
<section class="py-2 ">
<div class="custom-container text-center">
<p class="mb-4">
Evaluation sets include 20,000 images with over 66,000 reasoning annotations across 10 up-to-data generative models
and are balanced in terms of authenticity.<br>
Here we illustrate several real and fake images from each evaluation set.
</p>
<div id="mmfrEvaluationCarousel" class="carousel slide" data-bs-interval="false">
<div class="carousel-inner">
<div class="carousel-item active">
<h4 class="text-center">GPT-4o</h4>
<img loading="lazy" src="static/images/GPT-4o.png" class="responsive-image" alt="Evaluation Image 1">
</div>
<div class="carousel-item">
<h4 class="text-center">FLUX</h4>
<img loading="lazy" src="static/images/FLUX.png" class="responsive-image" alt="Evaluation Image 2">
</div>
<div class="carousel-item">
<h4 class="text-center">PixArt-α</h4>
<img loading="lazy" src="static/images/PixArt-α.png" class="responsive-image" alt="Evaluation Image 3">
</div>
<div class="carousel-item">
<h4 class="text-center">Kandinsky</h4>
<img loading="lazy" src="static/images/Kandinsky.png" class="responsive-image" alt="Evaluation Image 4">
</div>
<div class="carousel-item">
<h4 class="text-center">Midjourney</h4>
<img loading="lazy" src="static/images/Midjourney.png" class="responsive-image" alt="Evaluation Image 5">
</div>
<div class="carousel-item">
<h4 class="text-center">DeepFloyd IF</h4>
<img loading="lazy" src="static/images/DeepFloyd_IF.png" class="responsive-image" alt="Evaluation Image 6">
</div>
<div class="carousel-item">
<h4 class="text-center">DALLE-3</h4>
<img loading="lazy" src="static/images/DALLE-3.png" class="responsive-image" alt="Evaluation Image 7">
</div>
<div class="carousel-item">
<h4 class="text-center">Stable Diffusion</h4>
<img loading="lazy" src="static/images/Stable_Diffusion.png" class="responsive-image" alt="Evaluation Image 8">
</div>
<div class="carousel-item">
<h4 class="text-center">GigaGAN</h4>
<img loading="lazy" src="static/images/GigaGAN.png" class="responsive-image" alt="Evaluation Image 9">
</div>
<div class="carousel-item">
<h4 class="text-center">StyleGAN-XL</h4>
<img loading="lazy" src="static/images/StyleGAN-XL.png" class="responsive-image" alt="Evaluation Image 10">
</div>
</div>
<button class="carousel-control-prev" type="button" data-bs-target="#mmfrEvaluationCarousel"
data-bs-slide="prev">
<span class="carousel-control-prev-icon" aria-hidden="true"></span>
<span class="visually-hidden">Previous</span>
</button>
<button class="carousel-control-next" type="button" data-bs-target="#mmfrEvaluationCarousel"
data-bs-slide="next">
<span class="carousel-control-next-icon" aria-hidden="true"></span>
<span class="visually-hidden">Next</span>
</button>
</div>
</div>
</section>
<!-- Citation Section -->
<section class="py-5" style="background: #ffffff;">
<div class="container">
<div class="row justify-content-center">
<div class="col-lg-8">
<div class="card border-0" style="background: #ffffff;">
<div class="card-body p-4">
<h2 class="mb-2 text-center">Cite Our Work</h2>
<div class="citation-box p-4 rounded" style="background-color: #f8f9fa;">
<pre class="m-0" style="font-family: 'Roboto Mono', monospace; font-size: 0.9rem; color: #2d3748;">
@article{gao2025fakereasoning,
title={FakeReasoning: Towards Generalizable Forgery Detection and Reasoning},
author={Gao, Yueying and Chang, Dongliang and Yu, Bingyao and Qin, Haotian and Chen, Lei and Liang, Kongming and Ma, Zhanyu},
journal={arXiv preprint arXiv:2503.21210},
year={2025},
url={https://arxiv.org/abs/2503.21210}
}
</pre>
</div>
<div class="text-center mt-4">
<button class="btn btn-light mx-2 copy-btn border" data-clipboard-target=".citation-box pre">
<i class="bi bi-clipboard me-2"></i>Copy Citation
</button>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<!-- Footer -->
<footer>
<div class="py-5 bg-dark text-light text-center">
<p>This page was built upon the Academic Project Page Template.<br>Licensed under a <a
href="http://creativecommons.org/licenses/by-sa/4.0/" target="_blank" class="text-light">Creative Commons
Attribution-ShareAlike 4.0 License</a>.</p>
</div>
</footer>
<!-- Scripts -->
<script src="https://cdn.jsdelivr.net/npm/bootstrap@5.3.0/dist/js/bootstrap.bundle.min.js"></script>
<!-- <script src="https://unpkg.com/aos@2.3.1/dist/aos.js"></script> -->
<script src="https://unpkg.com/swiper/swiper-bundle.min.js"></script>
<!-- Clipboard.js -->
<script src="https://cdnjs.cloudflare.com/ajax/libs/clipboard.js/2.0.8/clipboard.min.js"></script>
<script>
// Initialize clipboard
new ClipboardJS('.copy-btn');
// Add feedback when copied
document.querySelectorAll('.copy-btn').forEach(btn => {
btn.addEventListener('click', function () {
const originalText = this.innerHTML;
this.innerHTML = '<i class="bi bi-check-circle me-2"></i>Copied';
setTimeout(() => {
this.innerHTML = originalText;
}, 2000);
});
});
</script>
<script>
function switchTab(event, tabId) {
document.querySelectorAll('.tab-button').forEach(btn => {
btn.classList.remove('active');
});
document.querySelectorAll('.tab-content').forEach(content => {
content.classList.remove('active');
});
event.currentTarget.classList.add('active');
document.getElementById(tabId).classList.add('active');
}
</script>
</body>
</html>