-
Notifications
You must be signed in to change notification settings - Fork 2
Expand file tree
/
Copy pathindex.html
More file actions
481 lines (467 loc) · 26.2 KB
/
index.html
File metadata and controls
481 lines (467 loc) · 26.2 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
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Wake Vision Dataset</title>
<link rel="stylesheet" href="styles.css">
</head>
<script>
document.addEventListener("DOMContentLoaded", () => {
const tables = document.querySelectorAll(".leaderboard-container table"); // Select all tables
tables.forEach(table => {
const headers = table.querySelectorAll("thead th");
const tbody = table.querySelector("tbody");
const defaultSortColumnIndex = 5; // Index of the "Test Accuracy" column
headers.forEach((header, index) => {
const arrow = document.createElement("span");
arrow.classList.add("arrow");
header.appendChild(arrow);
header.addEventListener("click", () => {
const isAscending = header.classList.toggle("sorted-asc");
header.classList.toggle("sorted-desc", !isAscending);
headers.forEach((h, i) => {
const hArrow = h.querySelector(".arrow");
if (i !== index) {
h.classList.remove("sorted-asc", "sorted-desc");
hArrow.textContent = "";
} else {
hArrow.textContent = isAscending ? "▲" : "▼";
}
});
const rows = Array.from(tbody.querySelectorAll("tr"));
const compare = (rowA, rowB) => {
const cellA = rowA.children[index].textContent.trim();
const cellB = rowB.children[index].textContent.trim();
const valueA = parseCellValue(cellA);
const valueB = parseCellValue(cellB);
if (valueA < valueB) return isAscending ? -1 : 1;
if (valueA > valueB) return isAscending ? 1 : -1;
return 0;
};
rows.sort(compare);
rows.forEach(row => tbody.appendChild(row));
});
});
const parseCellValue = (value) => {
if (value.includes('%')) {
return parseFloat(value.replace(/[^\d.-]/g, ''));
}
if (value.includes(',')) {
return parseFloat(value.replace(/,/g, ''));
}
return isNaN(value) ? value.toLowerCase() : parseFloat(value);
};
const defaultSortHeader = headers[defaultSortColumnIndex];
defaultSortHeader.classList.add("sorted-desc");
const rows = Array.from(tbody.querySelectorAll("tr"));
const compare = (rowA, rowB) => {
const cellA = rowA.children[defaultSortColumnIndex].textContent.trim();
const cellB = rowB.children[defaultSortColumnIndex].textContent.trim();
const valueA = parseCellValue(cellA);
const valueB = parseCellValue(cellB);
if (valueA < valueB) return 1;
if (valueA > valueB) return -1;
return 0;
};
rows.sort(compare);
rows.forEach(row => tbody.appendChild(row));
defaultSortHeader.querySelector(".arrow").textContent = "▼";
});
});
</script>
<body>
<header>
<div class="container">
<div class="header-content">
<div class="title-container">
<h1>Wake Vision Dataset</h1>
<div class="fade-in">
<a href="https://github.com/colbybanbury/Wake_Vision_Quickstart" class="button">Quick Start Guide</a>
<a href="https://arxiv.org/abs/2405.00892" class="button">Read the Paper</a>
<a href="#access" class="button">Access the Dataset</a>
</div>
</div>
<div class="logo-container">
<img src="wake_vision_logo.png" alt="Wake Vision Logo" class="logo">
</div>
</div>
</div>
</header>
<main class="container">
<h2 class="fade-in">About</h2>
<p class="fade-in">Wake Vision is a state-of-the-art person detection dataset specifically created for TinyML applications.
It provides a comprehensive collection of high-quality images and precise annotations to train and evaluate machine learning models for efficient person detection on embedded and edge devices.
Wake Vision also includes a fine-grain benchmark suite for evaluating the robustness of TinyML models.
</p>
<div class="sections-container">
<div class="section fade-in">
<h2>The Dataset</h2>
<p>Wake Vision is a large, high-quality binary image classifcation dataset for person detection:</p>
<ul>
<li>Over 6 million high-quality images</li>
<li>Two training sets (Large & Quality)</li>
<li>High quality validation and test sets (~2% Label Error Rate)</li>
</ul>
</div>
<div class="section fade-in">
<h2>Fine-Grain Benchmark Suite</h2>
<p>Wake Vision also incorporates a comprehensive fine-grained benchmark to assess fairness and robustness across:</p>
<ul>
<li>Perceived gender</li>
<li>Perceived age</li>
<li>Subject distance</li>
<li>Lighting conditions</li>
<li>Depictions (e.g., drawings, digital renderings)</li>
</ul>
</div>
</div>
<a class="anchor" id="access"></a>
<h2 class="fade-in">Access The Dataset</h2>
<div class="fade-in">
<a href="https://huggingface.co/datasets/Harvard-Edge/Wake-Vision" class="button">HuggingFace</a>
<a href="https://www.tensorflow.org/datasets/catalog/wake_vision" class="button">TensorFlow Datasets</a>
<a href="https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/1HOPXC" class="button">Harvard Dataverse</a>
<a href="https://github.com/harvard-edge/Wake_Vision/tree/main/download_with_python" class="button">Download Directly</a>
</div>
<h2 class="fade-in">Key Features</h2>
<div class="feature-grid">
<div class="feature-item fade-in">
<h3>TinyML Focus</h3>
<p>TinyML relevant usescase and tractable task.</p>
</div>
<div class="feature-item fade-in">
<h3>Two Training Sets</h3>
<p>One large and one high quality, ideal for data-centric AI research</p>
</div>
<div class="feature-item fade-in">
<h3>Diverse Scenarios</h3>
<p>Wide range of person detection use cases</p>
</div>
<div class="feature-item fade-in">
<h3>High-Quality Test and Val</h3>
<p>Manually labeled to ensure reliable evaluation</p>
</div>
</div>
<h2></h2>
<a class="anchor" id="modelzoo"></a>
<h2 class="fade-in">Model Zoo</h2>
<div class="leaderboard-wrapper fade-in">
<div class="leaderboard-container">
<table width="80%" style="margin: 0 auto; border:0px solid;text-align:center">
<thead>
<tr>
<th>Model Name</th>
<th>Input Size</th>
<th>RAM (KiB)</th>
<th>Flash (KiB)</th>
<th>MACs</th>
<th>Test Accuracy</th>
</tr>
</thead>
<tbody>
<tr>
<td align="center"><a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_mcunet-320kb-1mb_vww.tflite">mcunet-320kb</a></td>
<td align="center">(144,144,3)</td>
<td align="center">393</td>
<td align="center">923.76</td>
<td align="center">56,022,934</td>
<td align="center">85.9%</td><!--Accuracy for model zoo-->
<!--<td align="center">85.6±0.34%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_mobilenetv2.tflite">MobileNetV2_0.25</a></td>
<td align="center">(224,224,3)</td>
<td align="center">1,244.5</td>
<td align="center">410.55</td>
<td align="center">36,453,732</td>
<td align="center">84.7%</td><!--Accuracy for model zoo-->
<!--<td align="center">84.9±0.11%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_mcunet-5fps_vww.tflite">mcunet-5fps</a></td>
<td align="center">(80,80,3)</td>
<td align="center">226.5</td>
<td align="center">624.55</td>
<td align="center">11,645,502</td>
<td align="center">83.1%</td><!--Accuracy for model zoo-->
<!--<td align="center">82.9±0.29%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_mcunet-10fps_vww.tflite">mcunet-10fps</a></td>
<td align="center">(64,64,3)</td>
<td align="center">168.5</td>
<td align="center">533.84</td>
<td align="center">5,998,334</td>
<td align="center">82%</td><!--Accuracy for model zoo-->
<!--<td align="center">81.7±0.28%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_micronets_vww4_128_128_INT8.tflite">micronet_vww4</a></td>
<td align="center">(128,128,1)</td>
<td align="center">123.50</td>
<td align="center">417.03</td>
<td align="center">18,963,302</td>
<td align="center">78.6%</td><!--Accuracy for model zoo-->
<!--<td align="center">77.9±0.6%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_micronets_vww3_128_128_INT8.tflite">micronet_vww3</a></td>
<td align="center">(128,128,1)</td>
<td align="center">137.50</td>
<td align="center">463.73</td>
<td align="center">22,690,291</td>
<td align="center">78.5%</td><!--Accuracy for model zoo-->
<!--<td align="center">77.8±0.56%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_k_8_c_5.tflite">colabnas_k_8</a></td>
<td align="center">(50,50,3)</td>
<td align="center">32.5</td>
<td align="center">44.56</td>
<td align="center">2,135,476</td>
<td align="center">77.7%</td><!--Accuracy for model zoo-->
<!--<td align="center">77.3±0.37%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_k_4_c_5.tflite">colabnas_k_4</a></td>
<td align="center">(50,50,3)</td>
<td align="center">22</td>
<td align="center">18.49</td>
<td align="center">688,790</td>
<td align="center">75.9%</td><!--Accuracy for model zoo-->
<!--<td align="center">75.7±0.18%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_micronets_vww2_50_50_INT8.tflite">micronet_vww2</a></td>
<td align="center">(50,50,1)</td>
<td align="center">71.50</td>
<td align="center">225.54</td>
<td align="center">3,167,382</td>
<td align="center">72.5%</td><!--Accuracy for model zoo-->
<!--<td align="center">71.9±0.67%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_k_2_c_3.tflite">colabnas_k_2</a></td>
<td align="center">(50,50,3)</td>
<td align="center">18.5</td>
<td align="center">7.66</td>
<td align="center">250,256</td>
<td align="center">71.7%</td><!--Accuracy for model zoo-->
<!--<td align="center">70.6±0.96%</td>--><!--Numbers from paper-->
</tr>
</tbody>
</table>
</div>
</div>
</p>
<div class="feature-grid">
<div class="feature-grid">
<div class="feature-item fade-in">
<h3>🙋♂️ Contribute</h3>
<p>
<a href="mailto:AndreaMattia.Garavagno@santannapisa.it">Share your results with us</a>
and contribute to the leaderboard, or you can issue a PR at
<a href="https://github.com/harvard-edge/Wake_Vision_Webpage/pulls">this link</a>!
</p>
</div>
<div class="feature-item fade-in">
<h3>🏆 Challenge</h3>
<p>The first edition of the <a href="https://edgeai.modelnova.ai/contests/details/challenge-edge:-wake-vision">Wake Vision Challenge</a> has ended. Stay tuned for the next edition!</p>
</div>
</div>
</div>
<h2></h2>
<a class="anchor" id="challenge"></a>
<h2 class="fade-in">Wake Vision Challenge 🏆</h2>
<p class="fade-in">Following the release of the Wake Vision Dataset, the <a href="https://edgeai.modelnova.ai/contests/details/challenge-edge:-wake-vision">Wake Vision Challenge</a> was launched to advance research in TinyML. Participants were invited to contribute innovative model architectures in the model-centric track or to improve the dataset through the data-centric track. This section reports the results of the challenge. By clicking on the author name visitors can access the original submission containing the submitted model, source code, and a report describing the adopted solution. This provides a foundation for those interested in pushing the boundaries of TinyML with the Wake Vision Dataset.</p>
<h3 id="model-centric-track">Model-Centric Track</h3>
<div class="leaderboard-wrapper fade-in">
<div class="leaderboard-container">
<table width="80%" style="margin: 0 auto; border:0px solid;text-align:center">
<thead>
<tr>
<th>Model Name</th>
<th>Input Size</th>
<th>RAM (KiB)</th>
<th>Flash (KiB)</th>
<th>MACs</th>
<th>Test Accuracy</th>
</tr>
</thead>
<tbody>
<tr>
<td align="center"><a href="https://github.com/UtalTIPE/Wake_Vision_Challenge_Model_Centric_Track">ymac</a> (<a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_ymac.tflite">Model</a>)</td>
<td align="center">(50,50,3)</td>
<td align="center">34.5</td>
<td align="center">27.77</td>
<td align="center">431,985</td>
<td align="center">77.2%</td><!--Accuracy for model zoo-->
<!--<td align="center">76.7±0.51%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/samy101/Wake_Vision_Challenge_Model_Centric_Track">samy</a> (<a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_samy.tflite">Model</a>)</td>
<td align="center">(80,80,3)</td>
<td align="center">73.5</td>
<td align="center">34.55</td>
<td align="center">5,718,046</td>
<td align="center">79.9%</td><!--Accuracy for model zoo-->
<!--<td align="center">79.5±0.61%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/Anasb73/Wake_Vision_Challenge_Model_Centric_Track">anas-benalla</a> (<a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_anas_benalla.tflite">Model</a>)</td>
<td align="center">(50,50,3)</td>
<td align="center">48.5</td>
<td align="center">104.55</td>
<td align="center">4,899,292</td>
<td align="center">80%</td><!--Accuracy for model zoo-->
<!--<td align="center">79.7±0.28%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/Mohammad-Hallaq/Wake_Vision_Challenge_Model_Centric_Track">mohammad_hallaq</a> (<a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_mohammad_hallaq.tflite">Model</a>)</td>
<td align="center">(80,80,3)</td>
<td align="center">129.5</td>
<td align="center">128.32</td>
<td align="center">3,887,331</td>
<td align="center">77.8%</td><!--Accuracy for model zoo-->
<!--<td align="center">77.3±0.5%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/Elios-Lab/Wake_Vision_Challenge_Model_Centric_Track">apighetti</a> (<a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_apighetti.tflite">Model</a>)</td>
<td align="center">(50,50,3)</td>
<td align="center">48.5</td>
<td align="center">278.36</td>
<td align="center">693,818</td>
<td align="center">74.5%</td><!--Accuracy for model zoo-->
<!--<td align="center">74.3±0.22%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/benx13/Wake_Vision_Challenge_Model_Centric_Track">benx13</a> (<a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_benx13.tflite">Model</a>)</td>
<td align="center">(48,48,3)</td>
<td align="center">63.5</td>
<td align="center">94.95</td>
<td align="center">1,693,474</td>
<td align="center">79.9%</td><!--Accuracy for model zoo-->
<!--<td align="center">79.6±0.22%</td>--><!--Numbers from paper-->
</tr>
<tr>
<td align="center"><a href="https://github.com/Toledo1/Wake_Vision_Challenge_Model_Centric_Track.git">cezar</a> (<a href="https://github.com/harvard-edge/Wake_Vision/blob/main/model_zoo/wv_quality_cezar.tflite">Model</a>)</td>
<td align="center">(50,50,3)</td>
<td align="center">245.5</td>
<td align="center">57.23</td>
<td align="center">20,435,868</td>
<td align="center">79.5%</td><!--Accuracy for model zoo-->
<!--<td align="center">79.0±0.42%</td>--><!--Numbers from paper-->
</tr>
</tbody>
</table>
</div>
</div>
</p>
<h3 id="data-centric-track-2">Data-Centric Track - Edition 2</h3>
<div class="leaderboard-wrapper fade-in">
<div class="leaderboard-container">
<table width="80%" style="margin: 0 auto; border:0px solid;text-align:center">
<table width="80%" style="margin: 0 auto; border:0px solid;text-align:center"><thead>
<tr>
<th>Author</th>
<th>Test Accuracy</th>
</tr>
</thead>
<tbody>
<tr>
<td><a href="https://github.com/Anasb73/wake_vision_challenge_2_data_centric_track">Anasb73</a></td>
<td>84.5%</td>
</tr>
<tr>
<td><a href="https://github.com/M-A-Edwards/wake_vision_challenge_2_data_centric_track">M-A-Edwards</a></td>
<td>83.1%</td>
</tr>
<tr>
<td><a href="https://github.com/petriok/wake_vision_challenge_2_data_centric_track">petriok</a></td>
<td>77.9%</td>
</tr>
</tbody>
</table>
</div>
</div>
</p>
<h3 id="data-centric-track-1">Data-Centric Track - Edition 1</h3>
<div class="leaderboard-wrapper fade-in">
<div class="leaderboard-container">
<table width="80%" style="margin: 0 auto; border:0px solid;text-align:center">
<table width="80%" style="margin: 0 auto; border:0px solid;text-align:center"><thead>
<tr>
<th>Author</th>
<th>Test Accuracy</th>
</tr>
</thead>
<tbody>
<tr>
<td><a href="https://github.com/kais-bedioui/Wake_Vision_Challenge_Data_Centric_Track">kooks</a></td>
<td>79.2%</td>
</tr>
<tr>
<td><a href="https://github.com/rgroh1996/Wake_Vision_Challenge_Data_Centric_Track">rgroh</a></td>
<td>76.7%</td>
</tr>
<tr>
<td><a href="https://github.com/benx13/Wake_Vision_Challenge_Data_Centric_Track/tree/main">benx13</a></td>
<td>49.6%</td>
</tr>
</tbody>
</table>
</div>
</div>
<h2></h2>
<h2 class="fade-in">Example Images</h2>
<div class="image-grid">
<div class="image-item fade-in">
<img src="female_person.png" alt="Predominantly Female Person">
</div>
<div class="image-item fade-in">
<img src="bright_image.png" alt="Bright Image">
</div>
<div class="image-item fade-in">
<img src="depiction_person.png" alt="Depicted Person">
</div>
<div class="image-item fade-in">
<img src="young_person.png" alt="Young Person">
</div>
</div>
<h2></h2>
<h2 class="fade-in">License</h2>
<p class="fade-in">The Wake Vision labels are derived from Open Image's annotations which are licensed by Google LLC under CC BY 4.0 license. The images are listed as having a CC BY 2.0 license. Note from Open Images: "while we tried to identify images that are licensed under a Creative Commons Attribution license, we make no representations or warranties regarding the license status of each image and you should verify the license for each image yourself."</p>
<h2></h2>
<h2>Cite</h2>
<section id="cite" class="section">
@article{banbury2024wake,<br>
title={Wake Vision: A Tailored Dataset and Benchmark Suite for TinyML Computer Vision Applications},<br>
author={Banbury, Colby and Njor, Emil
and Garavagno, Andrea Mattia and
Stewart, Matthew and Warden, Pete
and Kudlur, Manjunath and Jeffries, Nat
and Fafoutis, Xenofon and Reddi, Vijay Janapa},<br>
journal={arXiv preprint arXiv:2405.00892},<br>
year={2024}<br>
}
</section>
<h2></h2>
<div class="contact">
<div>
<h2 class="fade-in">Contact</h2>
<p class="fade-in">For questions about the dataset or image takedown requests contact <a href="mailto:wakevision@edgeaifoundation.org">wakevision@edgeaifoundation.org</a>
</p>
</div>
<div class="logo-container">
<img src="Harvard_logo.png" alt="Harvard SEAS Logo" class="logo">
</div>
</div>
</main>
<footer>
<div class="container">
<p>© 2024 Wake Vision Dataset. All rights reserved.</p>
</div>
</footer>
</body>
</html>