-
Notifications
You must be signed in to change notification settings - Fork 4
Expand file tree
/
Copy pathindex.html
More file actions
890 lines (845 loc) · 40.4 KB
/
index.html
File metadata and controls
890 lines (845 loc) · 40.4 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
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
<!DOCTYPE html>
<html>
<head>
<title>EgoThink</title>
<style>
.hidden {
display: none;
}
</style>
<script src="https://cdn.jsdelivr.net/npm/chart.js"></script>
<meta charset="utf-8">
<meta name="description"
content="Evaluating First-Person Perspective Thinking Capability of Vision-Language Models">
<meta name="keywords" content="EgoThink, VLM, VLM Evaluation, Vision Language Model, First-Person, Egocentric, Agent, artificial intelligence, AI, AGI">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title> EgoThink: Evaluating First-Person Perspective Thinking Capability of Vision-Language Models</title>
<link rel="icon" href="./static/images/EgoThink_icon.png">
<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="stylesheet" href="./static/css/leaderboard.css">
<!-- <link href="https://unpkg.com/tabulator-tables@5.5.2/dist/css/tabulator_bulma.min.css" rel="stylesheet">
<script type="text/javascript" src="https://unpkg.com/tabulator-tables@5.5.2/dist/js/tabulator.min.js"></script> -->
<script type="text/javascript" src="static/js/sort-table.js" defer></script>
<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>
<script src="./static/js/question_card.js"></script>
<script src="./data/results/data_setting.js" defer></script>
<script src="./data/results/model_scores.js" defer></script>
<script src="./visualizer/data/data_public.js" defer></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 class="navbar-menu">
<div class="navbar-start" style="flex-grow: 1; justify-content: center;">
<div class="navbar-item has-dropdown is-hoverable">
<a class="navbar-link">
More Research
</a>
<div class="navbar-dropdown">
<!-- <a class="navbar-item" href="https://github.com/imoneoi/openchat">
<b>OpenChat</b> <p style="font-size:18px; display: inline; margin-left: 5px;">🔥</p>
</a> -->
<a class="navbar-item" href="https://adacheng.github.io/VidEgoThink/">
<b>VidEgoThink</b>
</a>
<a class="navbar-item" href="https://zhichengg.github.io/stb.github.io/">
<b>StableToolBench</b>
</a>
</div>
</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 is-bold">
<img src="static/images/EgoThink_icon.png" style="width:1em;vertical-align: middle" alt="Logo"/>
<span class="mmmu" style="vertical-align: middle">EgoThink</span>
</h1>
<h2 class="subtitle is-3 publication-subtitle">
Evaluating First-Person Perspective Thinking Capability of Vision-Language Models
</h2>
<div class="is-size-5 publication-authors">
<span class="author-block">Sijie Cheng*<sup style="color:#6fbf73;">†,1,2,3</sup>,</span>
<span class="author-block">Zhicheng Guo*<sup style="color:#ffac33;">1,2,3</sup>,</span>
<span class="author-block">Jingwen Wu*<sup style="color:#ed4b82;">4</sup>,</span>
<span class="author-block">Kechen Fang<sup style="color:#007bff;">5</sup>,</span><br>
<span class="author-block">Peng Li<sup style="color:#ed4b82;">✉,2</sup>,</span>
<span class="author-block">Huaping Liu<sup style="color:#ffac33;">1,3</sup>,</span>
<span class="author-block">Yang Liu<sup style="color:#ed4b82;">✉,1,2,3</sup>,</span>
<!-- <span class="author-block">Dongfu Jiang<sup style="color:#ffac33;">2</sup>,</span>
<span class="author-block">Weiming Ren<sup style="color:#ffac33;">2</sup>,</span>
<span class="author-block">Yuxuan Sun<sup style="color:#007bff;">4</sup>,</span>
<span class="author-block">Cong Wei<sup style="color:#ffac33;">2</sup>,</span>
<span class="author-block">Botao Yu<sup style="color:#ed4b82;">3</sup>,</span>
<span class="author-block">Ruibin Yuan<sup style="color:#ffac33;">5</sup>,</span>
<span class="author-block">Renliang Sun<sup style="color:#ffac33;">2</sup>,</span>
<span class="author-block">Ming Yin<sup style="color:#9b51e0;">7</sup>,</span>
<span class="author-block">Boyuan Zheng<sup style="color:#ed4b82;">3</sup>,</span>
<span class="author-block">Zhenzhu Yang<sup style="color:#007bff;">4</sup>,</span>
<span class="author-block">Yibo Liu<sup style="color:#ed4b82;">6</sup>,</span>
<span class="author-block">Wenhao Huang<sup style="color:#007bff;">4</sup>,</span><br>
<span class="author-block">Huan Sun*<sup style="color:#ed4b82;">3</sup>,</span>
<span class="author-block">Yu Su*<sup style="color:#ed4b82;">†,3</sup>,</span>
<span class="author-block">Wenhu Chen*<sup style="color:#ffac33;">†,2</sup></span> -->
</div>
<br>
<div class="is-size-5 publication-authors">
<span class="author-block"><sup style="color:#6fbf73;">1</sup>Department of Computer Science and Technology, Tsinghua University</span><br>
<span class="author-block"><sup style="color:#ffac33;">2</sup>Institute for AI Industry Research (AIR), Tsinghua University</span><br>
<span class="author-block"><sup style="color:#ed4b82;">3</sup>Beijing National Research Center for Information Science and Technology</span><br>
<span class="author-block"><sup style="color:#007bff;">4</sup>Department of Electrical and Computer Engineering, University of Toronto</span></br>
<span class="author-block"><sup style="color:#ffac33;">5</sup>Zhili College, Tsinghua University</span><br>
<!-- <span class="author-block"><sup style="color:#ed4b82;">6</sup>University of Victoria,</span>
<span class="author-block"><sup style="color:#9b51e0;">7</sup>Princeton University</span> -->
</div>
<br>
<div class="is-size-5 publication-authors">
<span class="author-block">*Equal contribution, ✉Corresponding author</span><br>
<span class="author-block">†Project Lead:</span>
<span class="author-block"><a href="mailto:csj23@mails.tsinghua.edu.cn">csj23@mails.tsinghua.edu.cn</a></span>
<!-- <span class="author-block"><a href="mailto:su.809@osu.edu">su.809@osu.edu</a>,</span>
<span class="author-block"><a href="mailto:wenhuchen@uwaterloo.ca">wenhuchen@uwaterloo.ca</a></span> -->
</div>
<div class="column has-text-centered">
<div class="publication-links">
<!-- PDF Link. -->
<span class="link-block">
<a href="https://arxiv.org/pdf/2311.15596.pdf"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>arXiv</span>
</a>
</span>
<span class="link-block">
<a href="https://huggingface.co/papers/2311.15596"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<p style="font-size:18px">🤗</p>
</span>
<span>HF Paper</span>
</a>
</span>
<span class="link-block">
<a href="https://huggingface.co/datasets/EgoThink/EgoThink"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<p style="font-size:18px">🤗</p>
</span>
<span>Dataset</span>
</a>
</span>
<span class="link-block">
<a href="https://github.com/AdaCheng/EgoThink"
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>
<!-- Dataset Link. -->
<!-- Leaderboard Link. -->
<span class="link-block">
<a href="#leaderboard"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<p style="font-size:18px">🏆</p>
</span>
<span>Leaderboard</span>
</a>
</span>
<!-- Visualization Link. -->
<!-- <span class="link-block">
<a href="#examples"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<p style="font-size:18px">📖</p>
</span>
<span>Examples</span>
</a>
</span> -->
<!-- Twitter Link. -->
<!-- <span class="link-block">
<a href="https://twitter.com/xiangyue96/status/1729698316554801358"
class="external-link button is-normal is-rounded is-dark">
<span class="icon"> -->
<!-- <i class="far fa-images"></i> -->
<!-- 💻🔗 -->
<!-- <p style="font-size:18px">🌐</p> -->
<!-- </span>
<span>Twitter</span>
</a>
</span> -->
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<style>
.center {
display: block;
margin-left: auto;
margin-right: auto;
width: 80%;
}
</style>
<section class="hero teaser">
<div class="container is-max-desktop">
<!-- <div class="hero-body">
<img src="static/images/tease_scores.png" alt="Examples from the dataset"/>
<h2 class="subtitle has-text-centered">
<span class="dnerf">Nerfies</span> turns selfie videos from your phone into
free-viewpoint
portraits.
</h2>
</div> -->
<!-- <div class="box m-5"> -->
<div class="content has-text-centered">
<img src="static/images/intro.png" alt="geometric reasoning" width="70%"/>
<p> <b><i>Figure 1:</i></b> The main categories of EgoThink to comprehensively assess the capability of thinking from a first-person perspective. </p>
</div>
<!-- </div> -->
<!-- </div> -->
</div>
</section>
<section class="section">
<div class="container" style="margin-bottom: 2vh;">
<!-- Abstract. -->
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">🔔News</h2>
<div class="content has-text-justified">
<p>
<b>[2024-10]: Our related paper <a href="https://arxiv.org/abs/2410.11623">VidEgoThink: Assessing Egocentric Video Understanding Capabilities for Embodied AI</a> has been released.</b><br>
<b>[2024-09]: EgoThink and VidEgoThink is invited to be presented in <a href="https://zhidx.com/p/441426.html">ZhiDX</a>.</b><br>
<b>[2024-04]: EgoThink is invited to be presented in ByteDance.</b><br>
<b>[2024-04]: EgoThink will be presented as a Poster (Highlight👀) in CVPR 2024.</b><br>
<b>[2024-03]: EgoThink is presented in <a href="https://www.bilibili.com/video/BV13D42157gP/?spm_id_from=333.1387.search.video_card.click">AITIME</a>. </b><br>
<b>[2024-02]: EgoThink has been accepted by CVPR 2024.</b><br>
<b>[2023-11]: Our paper <a href="https://arxiv.org/abs/2311.15596">Can Vision-Language Models Think from a First-Person Perspective?</a> has been released.</b><br>
<!-- <b>🔥[2023-11-27]: Our evaluation server for the test set is now available on <a href="https://eval.ai/web/challenges/challenge-page/2179/overview">EvalAI</a>. We welcome all submissions and look forward to your participation! 😆</b> -->
</p>
</div>
<h2 class="title is-3">Abstraction</h2>
<div class="content has-text-justified">
<p>
Vision-language models (VLMs) have recently shown promising results in traditional downstream tasks. Evaluation studies have emerged to assess their abilities, with the majority focusing on the third-person perspective, and only a few addressing specific tasks from the first-person perspective. However, the capability of VLMs to "think" from a first-person perspective, a crucial attribute for advancing autonomous agents and robotics, remains largely unexplored. To bridge this research gap, we introduce EgoThink, a novel visual question-answering benchmark that encompasses six core capabilities with twelve detailed dimensions. The benchmark is constructed using selected clips from egocentric videos, with manually annotated question-answer pairs containing first-person information. To comprehensively assess VLMs, we evaluate eighteen popular VLMs on EgoThink. Moreover, given the open-ended format of the answers, we use GPT-4 as the automatic judge to compute single-answer grading. Experimental results indicate that although GPT-4V leads in numerous dimensions, all evaluated VLMs still possess considerable potential for improvement in first-person perspective tasks. Meanwhile, enlarging the number of trainable parameters has the most significant impact on model performance on EgoThink. In conclusion, EgoThink serves as a valuable addition to existing evaluation benchmarks for VLMs, providing an indispensable resource for future research in the realm of embodied artificial intelligence and robotics.
</p>
</div>
</div>
</div>
<!--/ Abstract. -->
</div>
</section>
<!-- DATASET SECTION -->
<section class="hero is-light is-small">
<div class="hero-body has-text-centered">
<h1 class="title is-1 mmmu">
<img src="static/images/EgoThink_icon.png" style="width:1em;vertical-align: middle" alt="Logo"/>
<span class="mmmu" style="vertical-align: middle">EgoThink Benchmark</span>
</h1>
</div>
</section>
<!-- <section class="section">
<div class="container" style="margin-bottom: 2vh;">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Introduction</h2>
<div class="content has-text-justified">
<p> -->
<section class="section">
<div class="container">
<div class="columns is-centered has-text-centered">
<!-- <div class="column is-full-width has-text-centered"> -->
<div class="column is-four-fifths">
<h2 class="title is-3">Overview</h2>
<div class="content has-text-justified">
<p>
We specifically design six categories with twelve fine-grained dimensions from the first-person perspective for quantitative evaluation.
</p>
<img src="static/images/benchmark.jpg" alt="algebraic reasoning" class="center" width="110%">
<div class="content has-text-centered">
<p><b><i>Figure 2:</i></b> Categories with fine-grained dimensions and their corresponding examples of EgoThink benchmark.</p>
</div>
<p>
<ul>
<li>
<b>Object: What is around me?</b> Recognizing objects in the real world is a preliminary ability of the human visual system. Images from a first-person or egocentric perspective pay more attention to the objects surrounding the subject or in hands. Moreover, we further divide the object category into three fine-grained dimensions: (1) <i>Existence</i>, predicting whether there is an object as described in the images; (2) <i>Attribute</i>, detecting properties or characteristics (e.g., color) of an object;
(3) <i>Affordance</i>, predicting potential actions that a human can apply to an object.
</li>
<li>
<b>Activity: What am I doing?</b> Activity recognition is to automatically recognize specific human activities in video frames or still images. From the egocentric perspective, we mainly focus on actions or activities based on object-hand interaction.
</li>
<li>
<b>Localization: Where am I?</b> In reality, localization is a critical capability for navigation and scene understanding in the real world. Here we investigate the localization capability from two aspects, <i>Location</i> and <i>Spatial Relationship</i>. Location indicates detecting the scene surrounding the subject. Spatial reasoning contains allocentric and egocentric perspectives. We focus on the egocentric perspective, i.e., the position of the object with respect to the subject.
</li>
<li>
<b>Reasoning: What about the situation around me?</b> During the complex decision-making process, reasoning lies everywhere in our lives. Here we mainly focus on <i>Counting</i>, <i>Comparison</i>, and <i>Situated Reasoning</i>. Due to the first-person perspective, we generally count or compare objects in our hands or surrounding ourselves. As for situated reasoning, we employ cases that cannot be answered directly from the information in the images and require further reasoning processes.
</li>
<li>
<b>Forecasting: What will happen to me?</b> Forecasting is a critical skill in the real world. From an egocentric view, forecasting always predicts the future of object-state transformation or hand-object interactions.
</li>
<li>
<b>Planning: How will I do?</b> In reality, planning is an important capability to deal with complex problems, typically applied in <i>Navigation</i> and <i>Assistance</i>. Navigation is going to a goal location from a start position, while assistance is offering instructions to solve daily problems.
</li>
</ul>
</p>
</div>
</div>
</div>
<div class="columns is-centered has-text-centered">
<!-- <div class="column is-full-width has-text-centered"> -->
<div class="column is-four-fifths">
<h2 class="title is-3">Comparisons with Existing Benchmarks</h2>
<div class="content has-text-justified">
<p>
The ability to think from a first-person perspective is not adequately addressed by current evaluation benchmarks for VLMs. On one hand, most of these benchmarks (six out of nine, as listed in Table 1) focus solely on the third-person perspective. On the other hand, those benchmarks that do consider the first-person perspective only encompass a limited range of capabilities. For instance, EgoTaskQA examines spatial, temporal, and causal aspects, whereas EgoVQA is limited to object, action, and person aspects. Therefore, there is a clear need to develop a comprehensive benchmark to evaluate the first-person capabilities of VLMs more effectively.
</p>
<div class="content has-text-centered">
<img src="static/images/comparison.png" alt="algebraic reasoning" class="center">
<p><b><i>Table 1:</i></b> Comparison of recent comprehensive evaluation benchmarks of VLMs and our proposed benchmark EgoThink. Third and first indicate third-person and first-person perspectives. Datasets/Handcraft/LLMs denote existing datasets, manual annotation, and automatic generation by LLMs. PS/MC/OE indicate pairwise scoring, multi-choice, and open-ended question-answering, respectively.</p>
</div>
</div>
</div>
</div>
<div class="columns is-centered m-6">
<div class="column is-full has-text-centered content">
<h2 class="title is-3">Statistics</h2>
<!-- <div id="results-carousel" class="carousel results-carousel"> -->
<div class="content has-text-centered">
<img src="static/images/statistics.png" alt="algebraic reasoning" width="75%"/>
<p><b><i>Table 2:</i></b> Statistics of six categories with twelve dimensions in our EgoThink benchmark, where spatial* indicates spatial relationship and situated* indicates situated reasoning.</p>
</div>
<div class="content has-text-centered">
<img src="static/images/distribution.jpg" alt="arithmetic reasoning" width="35%"/>
<p><b><i>Figure 3:</i></b> This chart illustrates the distribution of various scene categories within the EgoThink dataset. The "others" category encompasses 13 different scene types, each representing less than one percent of total scenes.</p>
</div>
<!-- <div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/image_type_count.png" alt="arithmetic reasoning" width="80%"/>
<p> Distribution of image types in the MMMU dataset</p>
</div>
</div> -->
<!-- </div> -->
</div>
</div>
<!-- <div class="columns is-centered m-6">
<div class="column is-max-desktop has-text-centered">
<h2 class="title is-3" id="visualization">Visualization</h2>
<iframe src="visualizer/explore.html" style="width: 100%;min-height: 100vh; border-radius: 20px;"></iframe>
</div>
</div> -->
</div>
</section>
<!-- RESULTS SECTION -->
<section class="hero is-light is-small">
<div class="hero-body has-text-centered">
<h1 class="title is-1 mmmu" id="leaderboard">Leaderboard</h1>
</div>
</section>
<!-------------------------------------------------------------------- RESULTS SECTION -------------------------------------------------------------------->
<section class="section">
<div class="container">
<!-- <div class="columns is-centered has-text-centered"> -->
<!-- <div class="column is-full-width has-text-centered"> -->
<!-- <div class="column is-four-fifths"> -->
<!-- <h2 class="title is-3">Vision-Language Models</h2> -->
<!-- <div class="content has-text-centered"> -->
<!-- <img src="static/images/vlms.png" alt="algebraic reasoning" class="center"> -->
<!-- <p><b><i>Table 3:</i></b> Statistics of compared API-based and open-source VLMs, where TTP and ToP indicate Total Trainable Parameters and Total Parameters, respectively. Moreover, EgoData and Video indicate that there are egocentric visual data and video data for training, respectively.</p> -->
<!-- </div> -->
<!-- </div> -->
<!-- </div> -->
<div class="columns is-centered m-6">
<!-- <div class="column is-full has-text-centered content"> -->
<!-- <h2 class="title is-3" id="leaderboard">Leaderboard</h2> -->
<div class="content">
<div class="content has-text-justified">
<p>
Evaluating open-ended model generations is a non-trivial problem. To address this, we propose using GPT-4 as an automatic evaluator to better measure the generated answers. We continuously update the results of recent VLMs to ensure the effectiveness of EgoThink. Feel free to contribute to the performance of your model by adding it to our <a href="https://github.com/AdaCheng/EgoThink/blob/main/index.html">index.html</a>; we will review and merge it accordingly.
</p>
</div>
<!-- <button id="toggleButton" onclick="changeButtonText()">Switch to Validation Set</button> -->
<table id="table1" class="js-sort-table">
<tr>
<td class="js-sort-number"><strong>Reset</strong></td>
<td class="js-sort-number"><strong>Average</strong></td>
<td class="js-sort-number"><strong>Exist</strong></td>
<td class="js-sort-number"><strong>Attr</strong></td>
<td class="js-sort-number"><strong>Afford</strong></td>
<td class="js-sort-number"><strong>Activity</strong></td>
<td class="js-sort-number"><strong>Loc</strong></td>
<td class="js-sort-number"><strong>Spatial</strong></td>
<td class="js-sort-number"><strong>Count</strong></td>
<td class="js-sort-number"><strong>Compar</strong></td>
<td class="js-sort-number"><strong>Situtaed</strong></td>
<td class="js-sort-number"><strong>Forecasting</strong></td>
<td class="js-sort-number"><strong>Nav</strong></td>
<td class="js-sort-number"><strong>Assist</strong></td>
</tr>
<tr style="background-color: #f8fffe;">
<td style="text-align: left;"><b>GPT-4V(ision)</b></td>
<td><b>65.5</b></td>
<td>62.0</td>
<td><b>82.0</b></td>
<td><b>58.0</b></td>
<td><b>59.5</b></td>
<td style="text-decoration: underline;">86.0</td>
<td style="text-decoration: underline;">62.0</td>
<td><b>42.0</b></td>
<td>48.0</td>
<td><b>83.0</b></td>
<td><b>55.0</b></td>
<td><b>64.0</b></td>
<td><b>84.0</b></td>
</tr>
<tr style="background-color: #f9f2f8;">
<td style="text-align: left;"><b>OpenFlamingo-7B</b></td>
<td>27.2</td>
<td>16.0</td>
<td>55.0</td>
<td>37.0</td>
<td>15.0</td>
<td>34.0</td>
<td>34.0</td>
<td>21.0</td>
<td>40.0</td>
<td>21.0</td>
<td>31.0</td>
<td>11.0</td>
<td>11.0</td>
</tr>
<tr style="background-color: #f9f2f8;">
<td style="text-align: left;"><b>BLIP-2-6.7B</b></td>
<td>28.1</td>
<td>49.0</td>
<td>29.0</td>
<td>39.0</td>
<td>33.5</td>
<td>60.0</td>
<td>31.0</td>
<td>3.0</td>
<td>21.0</td>
<td>33.0</td>
<td>25.0</td>
<td>8.0</td>
<td>6.0</td>
</tr>
<tr style="background-color: #f9f2f8;">
<td style="text-align: left;"><b>LLaVA-1.5-7B</b></td>
<td>39.0</td>
<td>33.0</td>
<td>47.0</td>
<td style="text-decoration: underline;">54.0</td>
<td>35.5</td>
<td>35.0</td>
<td>49.0</td>
<td>20.0</td>
<td>47.0</td>
<td>37.0</td>
<td>27.0</td>
<td>29.0</td>
<td>54.0</td>
</tr>
<tr style="background-color: #f9f2f8;">
<td style="text-align: left;"><b>MiniGPT-4-7B</b></td>
<td>40.6</td>
<td>50.0</td>
<td>56.0</td>
<td>46.0</td>
<td>39.0</td>
<td>55.0</td>
<td>49.0</td>
<td>14.0</td>
<td>48.0</td>
<td>31.0</td>
<td>41.5</td>
<td>14.0</td>
<td>44.0</td>
</tr>
<tr style="background-color: #f9f2f8;">
<td style="text-align: left;"><b>InstructBLIP-7B</b></td>
<td>42.4</td>
<td>50.0</td>
<td>33.0</td>
<td>45.0</td>
<td>47.5</td>
<td>77.0</td>
<td>38.0</td>
<td>18.0</td>
<td>43.0</td>
<td>67.0</td>
<td>40.5</td>
<td>19.0</td>
<td>31.0</td>
</tr>
<tr style="background-color: #f9f2f8;">
<td style="text-align: left;"><b>LLaMA-Adapter-7B</b></td>
<td>42.5</td>
<td>37.0</td>
<td>60.0</td>
<td>46.0</td>
<td>34.5</td>
<td>48.0</td>
<td>51.0</td>
<td>29.0</td>
<td>39.0</td>
<td>25.0</td>
<td>41.5</td>
<td>42.0</td>
<td>57.0</td>
</tr>
<tr style="background-color: #f9f2f8;">
<td style="text-align: left;"><b>Otter-I-7B</b></td>
<td>45.3</td>
<td>48.0</td>
<td>56.0</td>
<td>39.0</td>
<td>44.0</td>
<td>60.0</td>
<td>44.0</td>
<td>39.0</td>
<td>48.0</td>
<td>42.0</td>
<td>38.0</td>
<td>31.0</td>
<td>55.0</td>
</tr>
<tr style="background-color: #f9f2f8;">
<td style="text-align: left;"><b>PandaGPT-7B</b></td>
<td>46.2</td>
<td>40.0</td>
<td>56.0</td>
<td>41.0</td>
<td>37.0</td>
<td>61.0</td>
<td>52.0</td>
<td>19.0</td>
<td style="text-decoration: underline;">52.0</td>
<td>53.0</td>
<td>43.0</td>
<td>39.0</td>
<td>61.0</td>
</tr>
<tr style="background-color: #f9f2f8;">
<td style="text-align: left;"><b>mPLUG-owl-7B</b></td>
<td>48.8</td>
<td>56.0</td>
<td>58.0</td>
<td>47.0</td>
<td>53.0</td>
<td>60.0</td>
<td>53.0</td>
<td>25.0</td>
<td>49.0</td>
<td>44.0</td>
<td>49.5</td>
<td>33.0</td>
<td>58.0</td>
</tr>
<tr style="background-color: #f9f2f8;">
<td style="text-align: left;"><b>LLaVA-7B</b></td>
<td>49.6</td>
<td>63.0</td>
<td>58.0</td>
<td>50.0</td>
<td>47.0</td>
<td>81.0</td>
<td>45.0</td>
<td>24.0</td>
<td>36.0</td>
<td>47.0</td>
<td>49.5</td>
<td>35.0</td>
<td>60.0</td>
</tr>
<tr style="background-color: #f4f9fe;">
<td style="text-align: left;"><b>InstructBLIP-13B</b></td>
<td>42.8</td>
<td>52.0</td>
<td>55.0</td>
<td>49.0</td>
<td>54.0</td>
<td>63.0</td>
<td>49.0</td>
<td>11.0</td>
<td>33.0</td>
<td>59.0</td>
<td>44.0</td>
<td>19.0</td>
<td>25.0</td>
</tr>
<tr style="background-color: #f4f9fe;">
<td style="text-align: left;"><b>PandaGPT-13B</b></td>
<td>43.1</td>
<td>35.0</td>
<td>52.0</td>
<td>41.0</td>
<td>40.5</td>
<td>68.0</td>
<td>31.0</td>
<td>32.0</td>
<td>40.0</td>
<td>47.0</td>
<td>45.5</td>
<td>16.0</td>
<td>69.0</td>
</tr>
<tr style="background-color: #f4f9fe;">
<td style="text-align: left;"><b>LLaVA-13B-Vicuna</b></td>
<td>46.4</td>
<td>54.0</td>
<td>62.0</td>
<td>52.0</td>
<td>46.0</td>
<td>53.0</td>
<td>46.0</td>
<td>26.0</td>
<td>44.0</td>
<td>29.0</td>
<td>44.0</td>
<td>35.0</td>
<td>66.0</td>
</tr>
<tr style="background-color: #f4f9fe;">
<td style="text-align: left;"><b>BLIP-2-11B</b></td>
<td>49.6</td>
<td>52.0</td>
<td>62.0</td>
<td>41.0</td>
<td>49.5</td>
<td><b>90.0</b></td>
<td><b>66.0</b></td>
<td>25.0</td>
<td>50.0</td>
<td>70.0</td>
<td>48.0</td>
<td>18.0</td>
<td>24.0</td>
</tr>
<tr style="background-color: #f4f9fe;">
<td style="text-align: left;"><b>InstructBLIP-11B</b></td>
<td>51.1</td>
<td><b>74.0</b></td>
<td style="text-decoration: underline;">68.0</td>
<td>48.0</td>
<td>49.5</td>
<td style="text-decoration: underline;">86.0</td>
<td>52.0</td>
<td>32.0</td>
<td>49.0</td>
<td style="text-decoration: underline;">73.0</td>
<td style="text-decoration: underline;">53.0</td>
<td>16.0</td>
<td>17.0</td>
</tr>
<tr style="background-color: #f4f9fe;">
<td style="text-align: left;"><b>LLaVA-13B-Llama2</b></td>
<td>55.1</td>
<td>65.0</td>
<td>61.0</td>
<td>45.0</td>
<td style="text-decoration: underline;">56.0</td>
<td>77.0</td>
<td>53.0</td>
<td style="text-decoration: underline;">34.0</td>
<td>34.0</td>
<td>66.0</td>
<td>50.5</td>
<td style="text-decoration: underline;">49.0</td>
<td style="text-decoration: underline;">71.0</td>
</tr>
<tr style="background-color: #f4f9fe;">
<td style="text-align: left;"><b>LLaVA-1.5-13B</b></td>
<td style="text-decoration: underline;">55.3</td>
<td style="text-decoration: underline;">66.0</td>
<td>55.0</td>
<td>51.0</td>
<td>55.0</td>
<td>82.0</td>
<td>57.0</td>
<td>32.0</td>
<td><b>56.0</b></td>
<td>67.0</td>
<td>48.5</td>
<td>39.0</td>
<td>55.0</td>
</tr>
</table>
<p> <b><i>Table 4:</i></b> Combined single-answer grading scores on zero-shot setups for various dimensions. The <b>bold</b> indicates the best performance while the <u>underline</u> indicates the second-best performance. Exist, Attr, Afford, Loc, Spatial, Count, Compar, Situated, Nav and Assist represent existence, attribute, affordance, location, spatial relationship, counting, comparison, situated reasoning, navigation, and assistance.</p>
</div>
<!-- </div> -->
</div>
<!-------------------------------------------------------------------- Correct Example -------------------------------------------------------------------->
<!-- <div class="columns is-centered m-6">
<div class="column is-full has-text-centered content">
<h2 class="title is-3" id="examples">Examples</h2>
<div id="results-carousel" class="carousel results-carousel">
<div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/case_1.png" alt="grade-lv" width="60%"/>
</div>
</div>
<div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/case_2.png" alt="grade-lv" width="60%"/>
</div>
</div>
<div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/case_3.png" alt="grade-lv" width="60%"/>
</div>
</div>
<div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/case_4.png" alt="grade-lv" width="60%"/>
</div>
</div>
<div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/case_5.png" alt="grade-lv" width="60%"/>
</div>
</div>
<div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/case_6.png" alt="grade-lv" width="60%"/>
</div>
</div>
<div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/case_7.png" alt="grade-lv" width="60%"/>
</div>
</div>
<div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/case_8.png" alt="grade-lv" width="60%"/>
</div>
</div>
<div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/case_9.png" alt="grade-lv" width="60%"/>
</div>
</div>
<div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/case_10.png" alt="grade-lv" width="60%"/>
</div>
</div>
<div class="box m-5">
<div class="content has-text-centered">
<img src="static/images/case_11.png" alt="grade-lv" width="60%"/>
</div>
</div>
</div>
</div>
</div> -->
</div>
</section>
<!-------------------------------------------------------------------- RESULTS SECTION -------------------------------------------------------------------->
<!-- @PAN TODO: bibtex -->
<section class="section" id="BibTeX">
<div class="container is-max-desktop content">
<h2 class="title is-3 has-text-centered">BibTeX</h2>
<pre><code>
@InProceedings{Cheng_2024_CVPR,
author = {Cheng, Sijie and Guo, Zhicheng and Wu, Jingwen and Fang, Kechen and Li, Peng and Liu, Huaping and Liu, Yang},
title = {EgoThink: Evaluating First-Person Perspective Thinking Capability of Vision-Language Models},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2024},
pages = {14291-14302}
}
</code></pre>
</div>
</section>
<footer class="footer">
<!-- <div class="container"> -->
<div class="content has-text-centered">
</div>
<div class="columns is-centered">
<div class="column is-8">
<div class="content has-text-centered">
<p>
This website is website adapted from <a href="https://mmmu-benchmark.github.io/">MMMU</a>, 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>
</div>
</div>
</div>
<!-- </div> -->
</footer>
<style>
.hidden {
display: none;
}
.sortable:hover {
cursor: pointer;
}
.asc::after {
content: ' ↑';
}
.desc::after {
content: ' ↓';
}
#toggleButton {
background-color: #ffffff;
border: 1px solid #dddddd;
color: #555555;
padding: 10px 20px;
text-align: center;
text-decoration: none;
display: inline-block;
font-size: 14px;
margin: 4px 2px;
cursor: pointer;
border-radius: 25px;
box-shadow: 0 4px 8px 0 rgba(0,0,0,0.2);
transition-duration: 0.4s;
}
#toggleButton:hover {
box-shadow: 0 12px 16px 0 rgba(0,0,0,0.24), 0 17px 50px 0 rgba(0,0,0,0.19); /* 鼠标悬停时的阴影效果 */
}
table {
border-collapse: collapse;
width: 100%;
margin-top: 5px;
border: 1px solid #ddd;
font-size: 14px;
}
th, td {
text-align: left;
padding: 8px;
}
th {
background-color: #f2f2f2;
border-bottom: 2px solid #ddd;
}
td:hover {background-color: #ffffff;}
</style>
</body>
</html>