|
| 1 | +<!DOCTYPE html> |
| 2 | +<meta charset="utf-8"> |
| 3 | + |
| 4 | +<html> |
| 5 | + |
| 6 | +<style type="text/css"> |
| 7 | +body { |
| 8 | + font-family: "Titillium Web", "HelveticaNeue-Light", "Helvetica Neue Light", "Helvetica Neue", Helvetica, Arial, "Lucida Grande", sans-serif; |
| 9 | + font-weight: 300; |
| 10 | + font-size: 17px; |
| 11 | + margin-left: auto; |
| 12 | + margin-right: auto; |
| 13 | + width: 980px; |
| 14 | +} |
| 15 | +h1 { |
| 16 | + font-weight:300; |
| 17 | + line-height: 1.15em; |
| 18 | +} |
| 19 | + |
| 20 | +h2 { |
| 21 | + font-size: 1.75em; |
| 22 | +} |
| 23 | +a:link,a:visited { |
| 24 | + color: #1367a7; |
| 25 | + text-decoration: none; |
| 26 | +} |
| 27 | +a:hover { |
| 28 | + color: #208799; |
| 29 | +} |
| 30 | +h1, h2, h3 { |
| 31 | + text-align: center; |
| 32 | +} |
| 33 | +h1 { |
| 34 | + font-size: 40px; |
| 35 | + font-weight: 500; |
| 36 | +} |
| 37 | +h2, h3 { |
| 38 | + font-weight: 400; |
| 39 | + margin: 16px 0px 4px 0px; |
| 40 | +} |
| 41 | +.paper-title { |
| 42 | + padding: 16px 0px 16px 0px; |
| 43 | +} |
| 44 | +section { |
| 45 | + margin: 32px 0px 32px 0px; |
| 46 | + text-align: justify; |
| 47 | + clear: both; |
| 48 | +} |
| 49 | +.col-5 { |
| 50 | + width: 20%; |
| 51 | + float: left; |
| 52 | +} |
| 53 | +.col-4 { |
| 54 | + width: 25%; |
| 55 | + float: left; |
| 56 | +} |
| 57 | +.col-2 { |
| 58 | + width: 50%; |
| 59 | + float: left; |
| 60 | +} |
| 61 | +.row, .author-row, .affil-row { |
| 62 | + overflow: auto; |
| 63 | +} |
| 64 | +.author-row, .affil-row { |
| 65 | + font-size: 20px; |
| 66 | +} |
| 67 | +.row { |
| 68 | + margin: 16px 0px 16px 0px; |
| 69 | +} |
| 70 | +.authors { |
| 71 | + font-size: 18px; |
| 72 | +} |
| 73 | +.affil-row { |
| 74 | + margin-top: 16px; |
| 75 | +} |
| 76 | +.teaser { |
| 77 | + max-width: 100%; |
| 78 | +} |
| 79 | +.text-center { |
| 80 | + text-align: center; |
| 81 | +} |
| 82 | +.screenshot { |
| 83 | + width: 256px; |
| 84 | + border: 1px solid #ddd; |
| 85 | +} |
| 86 | +.screenshot-el { |
| 87 | + margin-bottom: 16px; |
| 88 | +} |
| 89 | +hr { |
| 90 | + height: 1px; |
| 91 | + border: 0; |
| 92 | + border-top: 1px solid #ddd; |
| 93 | + margin: 0; |
| 94 | +} |
| 95 | +.material-icons { |
| 96 | + vertical-align: -6px; |
| 97 | +} |
| 98 | +p { |
| 99 | + line-height: 1.25em; |
| 100 | +} |
| 101 | +.caption_justify { |
| 102 | + font-size: 16px; |
| 103 | + color: #666; |
| 104 | + text-align: justify; |
| 105 | + margin-top: 0px; |
| 106 | + margin-bottom: 64px; |
| 107 | +} |
| 108 | +.caption { |
| 109 | + font-size: 16px; |
| 110 | + color: #666; |
| 111 | + text-align: center; |
| 112 | + margin-top: 8px; |
| 113 | + margin-bottom: 64px; |
| 114 | +} |
| 115 | +.caption_inline { |
| 116 | + font-size: 16px; |
| 117 | + color: #666; |
| 118 | + text-align: center; |
| 119 | + margin-top: 8px; |
| 120 | + margin-bottom: 0px; |
| 121 | +} |
| 122 | +.caption_bold { |
| 123 | + font-size: 16px; |
| 124 | + color: #666; |
| 125 | + text-align: center; |
| 126 | + margin-top: 0px; |
| 127 | + margin-bottom: 0px; |
| 128 | + font-weight: bold; |
| 129 | +} |
| 130 | +video { |
| 131 | + display: block; |
| 132 | + margin: auto; |
| 133 | +} |
| 134 | +figure { |
| 135 | + display: block; |
| 136 | + margin: auto; |
| 137 | + margin-top: 10px; |
| 138 | + margin-bottom: 10px; |
| 139 | +} |
| 140 | +#bibtex pre { |
| 141 | + font-size: 14px; |
| 142 | + background-color: #eee; |
| 143 | + padding: 16px; |
| 144 | +} |
| 145 | +.blue { |
| 146 | + color: #2c82c9; |
| 147 | + font-weight: bold; |
| 148 | +} |
| 149 | +.orange { |
| 150 | + color: #d35400; |
| 151 | + font-weight: bold; |
| 152 | +} |
| 153 | +.flex-row { |
| 154 | + display: flex; |
| 155 | + flex-flow: row wrap; |
| 156 | + justify-content: space-around; |
| 157 | + padding: 0; |
| 158 | + margin: 0; |
| 159 | + list-style: none; |
| 160 | +} |
| 161 | +.paper-btn { |
| 162 | + position: relative; |
| 163 | + text-align: center; |
| 164 | + |
| 165 | + display: inline-block; |
| 166 | + margin: 8px; |
| 167 | + padding: 8px 8px; |
| 168 | + |
| 169 | + border-width: 0; |
| 170 | + outline: none; |
| 171 | + border-radius: 2px; |
| 172 | + |
| 173 | + background-color: #1367a7; |
| 174 | + color: #ecf0f1 !important; |
| 175 | + font-size: 20px; |
| 176 | + width: 100px; |
| 177 | + font-weight: 600; |
| 178 | +} |
| 179 | +.paper-btn-parent { |
| 180 | + display: flex; |
| 181 | + justify-content: center; |
| 182 | + margin: 16px 0px; |
| 183 | +} |
| 184 | +.paper-btn:hover { |
| 185 | + opacity: 0.85; |
| 186 | +} |
| 187 | +.container { |
| 188 | + margin-left: auto; |
| 189 | + margin-right: auto; |
| 190 | + padding-left: 16px; |
| 191 | + padding-right: 16px; |
| 192 | +} |
| 193 | +.venue { |
| 194 | + color: #1367a7; |
| 195 | +} |
| 196 | +</style> |
| 197 | +<link href='https://fonts.googleapis.com/css?family=Titillium+Web:400,600,400italic,600italic,300,300italic' rel='stylesheet' type='text/css'> |
| 198 | +<head> |
| 199 | + <title>Assembly101: A Large-Scale Multi-View Video Dataset for Understanding Procedural Activities</title> |
| 200 | + <meta property="og:description" content="Assembly101: A Large-Scale Multi-View Video Dataset for Understanding Procedural Activities"/> |
| 201 | + <link href="https://fonts.googleapis.com/css2?family=Material+Icons" rel="stylesheet"> |
| 202 | +</head> |
| 203 | + |
| 204 | +<body> |
| 205 | +<div class="container"> |
| 206 | + <div class="paper-title"> |
| 207 | + <h1>Assembly101: A Large-Scale Multi-View Video Dataset for Understanding Procedural Activities</h1> |
| 208 | + </div> |
| 209 | + |
| 210 | + <div id="authors"> |
| 211 | + <div class="author-row"> |
| 212 | + <div class="col-4 text-center"><a href="https://scholar.google.com/citations?user=-juoweoAAAAJ">Fadime Sener<sup>1</sup></a></div> |
| 213 | + <div class="col-4 text-center"><a href="">Dibyadip Chatterjee<sup>2</sup></a></div> |
| 214 | + <div class="col-4 text-center"><a href="">Daniel Shelepov<sup>1</sup></a></div> |
| 215 | + <div class="col-4 text-center"><a href="">Kun He<sup>1</sup></a></div> |
| 216 | + <div class="col-4 text-center"><a href="">Dipika Singhania<sup>2</sup></a></div> |
| 217 | + <div class="col-4 text-center"><a href="">Robert Wang<sup>1</sup></a></div> |
| 218 | + <div class="col-4 text-center"><a href="">Angela Yao<sup>2</sup></a></div> |
| 219 | + </div> |
| 220 | + |
| 221 | + <div class="affil-row"> |
| 222 | + <div class="col-1 text-center"><sup>1</sup>Reality Labs at Meta</div> |
| 223 | + <div class="col-1 text-center"><sup>2</sup>National University of Singapore</div> |
| 224 | + </div> |
| 225 | + |
| 226 | + <div style="clear: both"> |
| 227 | + <div class="paper-btn-parent"> |
| 228 | + <a class="paper-btn" href="assets/Assembly101.pdf"> |
| 229 | + <span class="material-icons"> description </span> |
| 230 | + Paper |
| 231 | + </a> |
| 232 | + <a class="paper-btn" href="https://drive.google.com/drive/folders/1nh8PHwEw04zxkkkKlfm4fsR3IPEDvLKj?usp=sharing"> |
| 233 | + <span class="material-icons"> videocam </span> |
| 234 | + Dataset |
| 235 | + </a> |
| 236 | + <a class="paper-btn" href="https://github.com/assembly101/assembly101"> |
| 237 | + <span class="material-icons"> code </span> |
| 238 | + Code |
| 239 | + </a> |
| 240 | + </div> |
| 241 | + </div> |
| 242 | + </div> |
| 243 | + |
| 244 | + <section id="teaser-videos"> |
| 245 | + |
| 246 | + <figure style="width: 100%; float: left"> |
| 247 | + <video class="centered" width="100%" autoplay muted loop playsinline> |
| 248 | + <source src="assets/12_view_assembly.mp4" type="video/mp4"> |
| 249 | + Your browser does not support the video tag. |
| 250 | + </video> |
| 251 | + </figure> |
| 252 | + |
| 253 | + <figure style="width: 100%; float: left"> |
| 254 | + <p class="caption_justify"> |
| 255 | + Assembly101 is a large-scale video dataset for action recognition and markerless motion capture of hand-object interactions, captured in the above cage setting. |
| 256 | + The recordings, which are multi-view captures, feature participants assembling 101 children's toys. |
| 257 | + </p> |
| 258 | + </figure> |
| 259 | + </section> |
| 260 | + |
| 261 | + <section id="news"> |
| 262 | + <h2>News</h2> |
| 263 | + <hr> |
| 264 | + <div class="row"> |
| 265 | + <div><span class="material-icons"> description </span> [March 28th 2022] Paper released on <a href="https://arxiv.org/pdf/2203.14712.pdf">arXiv</a>.</div> |
| 266 | + <div><span class="material-icons"> integration_instructions </span> [March 28th 2022] Dataset released on <a href="https://drive.google.com/drive/folders/1nh8PHwEw04zxkkkKlfm4fsR3IPEDvLKj?usp=sharing">Google Drive</a>.</div> |
| 267 | + </div> |
| 268 | + </section> |
| 269 | + |
| 270 | + <section id="abstract"/> |
| 271 | + <h2>Abstract</h2> |
| 272 | + <hr> |
| 273 | + <p> |
| 274 | + Assembly101 is a new procedural activity dataset featuring 4321 videos of people assembling and disassembling 101 "take-apart" toy vehicles. Participants work without fixed instructions, and the sequences feature rich and natural variations in action ordering, mistakes, and corrections. Assembly101 is the first multi-view action dataset, with simultaneous static (8) and egocentric (4) recordings. Sequences are annotated with more than 100K coarse and 1M fine-grained action segments, and 18M 3D hand poses. We benchmark on three action understanding tasks: recognition, anticipation and temporal segmentation. Additionally, we propose a novel task of detecting mistakes. The unique recording format and rich set of annotations allow us to investigate generalization to new toys, cross-view transfer, long-tailed distributions, and pose vs. appearance. We envision that Assembly101 will serve as a new challenge to investigate various activity understanding problems. |
| 275 | + </p> |
| 276 | + </section> |
| 277 | + |
| 278 | + <section id="paper"> |
| 279 | + <h2>Paper</h2> |
| 280 | + <hr> |
| 281 | + <div class="flex-row"> |
| 282 | + <div style="box-sizing: border-box; padding: 16px; margin: auto;"> |
| 283 | + <a href="assets/paper.pdf"><img class="screenshot" src="assets/assembly101_rig.png"></a> |
| 284 | + </div> |
| 285 | + <div style="width: 60%"> |
| 286 | + <p><b>Assembly101: A Large-Scale Multi-View Video Dataset for Understanding Procedural Activities</b></p> |
| 287 | + <p>Fadime Sener, Dibyadip Chatterjee, Daniel Shelepov, Kun He, Dipika Singhania, Robert Wang, Angela Yao</p> |
| 288 | + |
| 289 | + <div><span class="material-icons"> description </span><a href="assets/Assembly101.pdf"> Paper preprint </a></div> |
| 290 | + <div><span class="material-icons"> description </span><a href="assets/Assembly101_supp.pdf"> Supplementary </a></div> |
| 291 | + <div><span class="material-icons"> description </span><a href="https://arxiv.org/pdf/2203.14712.pdf"> arXiv version</a></div> |
| 292 | + <div><span class="material-icons"> integration_instructions </span><a href="https://github.com/assembly101/assembly101"> Code</a></div> |
| 293 | +<!-- <div><span class="material-icons"> videocam </span><a href="https://youtu.be/asdf"> Video</a></div>--> |
| 294 | + |
| 295 | + <p>Please send feedback and questions to <tt>3dassembly101<at>gmail.com</tt></p> |
| 296 | + </div> |
| 297 | + </div> |
| 298 | + </section> |
| 299 | + |
| 300 | + <section id="bibtex"> |
| 301 | + <h2>Citation</h2> |
| 302 | + <hr> |
| 303 | + <pre><code>@article{sener2022assembly101, |
| 304 | + title = {Assembly101: A Large-Scale Multi-View Video Dataset for Understanding Procedural Activities}, |
| 305 | + author = {F. Sener and D. Chatterjee and D. Shelepov and K. He and D. Singhania and R. Wang and A. Yao}, |
| 306 | + journal = {CVPR 2022}, |
| 307 | +} |
| 308 | +</code></pre> |
| 309 | + </section> |
| 310 | +</div> |
| 311 | +</body> |
| 312 | +</html> |
0 commit comments