circle_dataset/index.html at main · Stanford-TML/circle_dataset · GitHub

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
<!doctype html>
<html lang="en">

<head>
  <meta charset="utf-8">
  <meta name="viewport" content="width=device-width, initial-scale=1">
  <title>CIRCLE: Capture in Rich Contextual Environments</title>
  <link href="//netdna.bootstrapcdn.com/bootstrap/3.2.0/css/bootstrap.min.css" rel="stylesheet">
  <link href="https://cdn.jsdelivr.net/npm/bootstrap@5.2.1/dist/css/bootstrap.min.css" rel="stylesheet"
    integrity="sha384-iYQeCzEYFbKjA/T2uDLTpkwGzCiq6soy8tYaI1GyVh/UjpbCx/TYkiZhlZB6+fzT" crossorigin="anonymous">

    <link rel="stylesheet" href="styles.css">
  <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=Montserrat:wght@500&display=swap" rel="stylesheet">


</head>


<body>
  <h1 class="project-title">CIRCLE: Capture in Rich Contextual Environments</h1>
  <!-- <img class="project-image" src="assets/placeholder.jpeg" alt=""> -->
  <div id="carousel-id" class="carousel" data-interval="false" style="width: 90%; margin: 0 auto; margin-top: 40px;">
    <ol class="carousel-indicators">
      <li data-target="#carousel-id" data-slide-to="0" class="active"></li>
      <li data-target="#carousel-id" data-slide-to="1" class=""></li>
      <li data-target="#carousel-id" data-slide-to="2" class=""></li>
      <li data-target="#carousel-id" data-slide-to="3" class=""></li>
    </ol>
    <div class="carousel-inner">
      <div class="item active">
        <div class="row" style="justify-content: row;">
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/smpl4.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/fpov4.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/dpov4.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
          </div>
      </div>
      <div class="item">
        <div class="row" style="justify-content: row;">
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/smpl2.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/fpov2.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/dpov2.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
          </div>
      </div>
      <div class="item">
        <div class="row" style="justify-content: row;">
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/smpl3.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/fpov3.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/dpov3.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
          </div>
      </div>
      <div class="item">
        <div class="row" style="justify-content: row;">
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/smpl1.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/fpov1.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/carousel/dpov1.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
          </div>
      </div>
    </div>
    <a class="left carousel-control" href="#carousel-id" data-slide="prev"><span class="glyphicon glyphicon-chevron-left"></span></a>
    <a class="right carousel-control" href="#carousel-id" data-slide="next"><span class="glyphicon glyphicon-chevron-right"></span></a>
  </div>
  <p class="caption">We introduce CIRCLE, a dataset of 3D whole body reaching motion and corresponding first person RGDB videos within a fully furnished virtual home.
  </p>


  <div class="authors">
    <a href="">Joao Pedro Araujo<sup>1</sup></a>
    <a href="">Jiaman Li<sup>1</sup></a>
    <a href="">Karthik Vetrivel<sup>1</sup></a>
    <a href="">Rishi Agarwal<sup>1</sup></a>
    <a href="">Deepak Gopinath<sup>2</sup></a>
    <a href="">Jiajun Wu<sup>1</sup></a>
    <a href="">Alexander Clegg<sup>2</sup></a>
    <a href="">C. Karen Liu<sup>1</sup></a>
  </div>

  <div class="affiliations">
    <span><sup>1</sup>Stanford University</span>
    <span><sup>2</sup>Meta AI</span>
  </div>

  <div class="project-icons">
    <a href="https://arxiv.org/abs/2303.17912">
      <img class="icon" src="assets/file-solid.svg" alt="">
      Paper
    </a>
    <!-- <a href="">
      <img class="icon" src="assets/github.svg" alt="">
      Code
    </a> -->

    <a href="https://github.com/Stanford-TML/circle_dataset">
      <img class="icon" src="assets/download-solid.svg" alt="">
      Dataset
    </a>
  </div>

  <section>
    <div class="abstract">
      <p>
        <h2 class="section-header ">Abstract</h2>
        Synthesizing 3D human motion in a contextual, ecological environment is important for simulating realistic activities people perform in the real world. However, conventional optics-based motion capture systems are not suited for simultaneously capturing human movements and complex scenes. The lack of rich contextual 3D human motion datasets presents a roadblock to creating high-quality generative human motion models. We propose a novel motion acquisition system in which the actor perceives and operates in a highly contextual virtual world while being motion captured in the real world. Our system enables rapid collection of high-quality human motion in highly diverse scenes, without the concern of occlusion or the need for physical scene construction in the real world. We present CIRCLE, a dataset containing 10 hours of full-body reaching motion from 5 subjects across nine scenes, paired with ego-centric information of the environment represented in various forms, such as RGBD videos. We use this dataset to train a model that generates human motion conditioned on scene information. Leveraging our dataset, the model learns to use ego-centric scene information to achieve nontrivial reaching tasks in the context of complex 3D scenes.
      </p>
    </div>
    <div class="teaser-images">
      <div>
        <img class="project-image" src="assets/teaser2.png" alt="">
        <p>Example poses from CIRCLE captured from real human motion in a virtual environment.</p>
      </div>
      <!-- <div>
        <img class="project-image" src="assets/teaser_back.png" alt="">
        <p>hi</p>
      </div> -->
    </div>

  </section>

  <section>
    <div class="abstract">
      <h2 class="section-header ">CIRCLE at a Glance</h2>
      <div class="numbers">
        <div>
          <h4>10</h4>
          <p>hours of data</p>
        </div>

        <div>
          <h4>5</h4>
          <p>subjects</p>
        </div>

        <div>
          <h4>9</h4>
          <p>scenes</p>
        </div>

        <div>
          <h4>7000+</h4>
          <p>sequences</p>
        </div>
      </div>
    </div>
    <p class="abstract" style>After post-processing, each sequence in CIRCLE include the <a href="https://smpl-x.is.tue.mpg.de/">SMPL-X</a> parameters of a body model fit to the mocap data using <a href="https://mosh.is.tue.mpg.de/">MoSh</a>, the VR headset trajectory synchronized to the mocap skeleton,
      egocentric RGB-D video (rendered with <a href="https://research.facebook.com/publications/habitat-a-platform-for-embodied-ai-research/">Habitat</a> and <a href="https://www.blender.org/">Blender</a>), task specific data, such as initial and goal conditions,
      and the scene where the data are collected.</p>
    </div>

  </section>

  <section>
    <div class="abstract">
      <h2 class="section-header ">Collection System</h2>
      <p>
        Our system enables collection of high-quality human motion in highly diverse scenes, without the concern of
        occlusion or the need for physical scene construction. For the purposes of this dataset, we used the Oculus
        Quest 2--though, our system is hardware agnostic.
      </p>

      <div class="image-grid" style="margin-top: 40px;">
        <div class="row">
          <h4>Real World</h4>
          <img src="assets/collection_system/image15.png" alt="">
          <img src="assets/collection_system/image16.png" alt="">
        </div>
        <div class="row">
          <h4>Virtual World</h4>
          <img src="assets/collection_system/image14.png" alt="">
          <img src="assets/collection_system/image17.png" alt="">
        </div>
        <div class="row">
          <h4>Egocentric View</h4>
          <img src="assets/collection_system/image18.png" alt="">
          <img src="assets/collection_system/image19.png" alt="">
        </div>
      </div>
  </section>

  <section class="section-gray">
    <div class="abstract">
      <h2 class="section-header">Wide Variety of Full-Body Motion Data</h2>
      <p>
        CIRCLE sequences range in difficulty, from easy sequences with no obstacles to sequences that require
        considerable full-body adjustment. The videos below shows the live-capture skeleton and scene rendered within the headset while the subject completes tasks.
      </p>
    </div>
      <div id="difficulty" style="display: flex; justify-content: center; margin-top: 40px;">
        <div class="result-video" style="width: 70%">
          <!-- Easy difficulty  -->
          <div class="row">
            <h4>Easy</h4>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/difficulty/easy1.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/difficulty/easy2.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
          </div>


          <!-- Medium difficulty  -->
          <div class="row">
            <h4>Medium</h4>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/difficulty/medium1.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
            <video loop muted>
              <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/difficulty/medium2.mp4" type="video/mp4">
              Your browser does not support the video tag.
            </video>
          </div>

           <!-- Hard difficulty  -->
      <div class="row">
        <h4>Hard</h4>
        <video loop muted>
          <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/difficulty/hard1.mp4" type="video/mp4">
          Your browser does not support the video tag.
        </video>
        <video loop muted>
          <source src="https://circledataset.s3.us-west-2.amazonaws.com/website/videos/difficulty/hard2.mp4" type="video/mp4">
          Your browser does not support the video tag.
        </video>
      </div>
    </div>


  </section>

  <section>
    <div class="abstract">
      <h2 class="section-header">Model</h2>
      <img style="width: 100%" src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/image41.png" alt="">
      <p style="margin-top: 20px;">Given a start pose, goal position, and 3D scene (a), we first initialize the input poses using constant local
        joint rotation and linearly interpolated root translation (b). We extract scene features for each time step using BPS or PointNet++ from the
        initialized poses, a fixed point set sampled from a cylinder, and scene point clouds (c). We then concatenate scene features and initialized
        poses and feed them to a transformer-based model (d) to generate final poses (e).</p>
      <h4 style="text-align: left; margin-top: 50px;">Model Input</h4>
      <p>
        We initialize the input to our model using a constant pose that linearly interpolates the start and goal wrist
        positions. The output is the edited motion.
      </p>
      <div style="display: flex; justify-content: space-between">
        <h4></h4>
        <h4>Input</h4>
        <h4></h4>
        <h4>Output</h4>
        <h4></h4>
        <h4>Ground Truth</h4>
        <h4></h4>
      </div>
      <video  loop muted style="width: 100%">
        <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media30.mp4.webm" type="video/webm">
        Your browser does not support the video tag.
      </video>


      <h4 style="text-align: left; margin-top: 50px;">Model Results</h4>
      <p>We compare our approach against two baselines, <a href="https://goal.is.tue.mpg.de/">GOAL</a> and NO-SCENE. <a href="https://goal.is.tue.mpg.de/">GOAL</a> is an MLP architecture that, given a
        start and goal pose, autoregressively predicts the next pose in the sequence. NO-SCENE employs our same
        architecture without scene encoding.</p>
    </div>
    <div style="display: flex; justify-content: center; margin-top: 40px;">
      <div class="result-video">
        <div class="row">
          <!-- Our Model, BPS -->
          <h4>Our Model (BPS)</h4>
          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media32.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>

          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media36.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>

          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media40.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>
        </div>

        <div class="model-gap">
        </div>

        <!-- No Scene -->
        <div class="row">
          <h4>No Scene    </h4>
          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media31.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>
          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media35.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>

          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media39.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>
        </div>


        <!-- GOAL -->
        <div class="row">
          <h4>GOAL</h4>
          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media33.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>
          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media37.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>

          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media41.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>
        </div>


        <!-- Ground Truth -->
        <div class="row">
          <h4>Ground Truth</h4>
          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media5.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>
          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media7.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>
          <video loop muted>
            <source src="https://circledemos.s3.us-west-2.amazonaws.com/model/model/media11.mp4.webm" type="video/webm">
            Your browser does not support the video tag.
          </video>
        </div>
      </div>
    </div>
  </section>


  </div>


  <!-- jQuery -->
  <script src="//code.jquery.com/jquery.js"></script>
  <!-- Bootstrap JavaScript -->
  <script src="//netdna.bootstrapcdn.com/bootstrap/3.2.0/js/bootstrap.min.js"></script>
  <script>
    // This script makes sure the carousel videos are in sync.

    const videos = document.getElementsByTagName('video');

    function processRow(row) {
      // Get the duration of the shortest video, and add an
      // event listener to the first video that restart all videos
      // in this row when it is time.
      const videos = row.querySelectorAll('video');
      let shortestDuration = Infinity;

      const firstVideo = videos[0];
      const otherVideos = Array.from(videos).slice(1);

      videos.forEach((video) => {
        if (video.duration < shortestDuration) {
          shortestDuration = video.duration;
        }
      });

      firstVideo.addEventListener('timeupdate', function() {
        if (this.currentTime > 0.8 * shortestDuration) {
          videos.forEach((video) => {
            video.currentTime = 0;
          });
        }
      });
    }

    function syncVideos() {
      // Make sure no videos are playing
      for (let i = 0; i < videos.length; i++) {
        videos[i].pause();
        videos[i].currentTime = 0;
      }

      // Wire the carousel videos so that when the shortest video of a
      // row finishes, all go back to the start
      const carouselInner = document.querySelector('.carousel-inner');
      const rows = carouselInner.querySelectorAll('.row');
      rows.forEach((row) => processRow(row));

      // Play the videos. The timeout seems to help...
      setTimeout(() => {
        for (let i = 0; i < videos.length; i++) {
          videos[i].play();
        }
      }, 1000);
    }

    function syncVideosWrapper() {
      let flag = false;

      // Check if all the videos have loaded already. If not,
      // call this function again in 1 second.
      for (let i = 0; i < videos.length; i++) {
        const video = videos[i];
        if (isNaN(video.duration)) {
          flag = true;
        }
      }

      if (flag) {
        setTimeout(() => {
          syncVideosWrapper();
        }, 1000);
      } else {
        syncVideos();
      }
    }

    syncVideosWrapper();
  </script>
  <script>
    // This script trims the difficulty videos
    const difficultyDiv = document.getElementById('difficulty');
    const difficutlyVideos = difficultyDiv.getElementsByTagName('video');

    Array.from(difficutlyVideos).forEach((video) => {
      video.addEventListener('timeupdate', function() {
        if (this.currentTime > 0.7 * this.duration) {
          this.currentTime = 0;
        }
      });
    });
  </script>
</body>

</html>