index.html 25.4 KB
Newer Older
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
<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
  <meta charset="utf-8">
  <meta http-equiv="X-UA-Compatible" content="IE=edge">
  <meta name="viewport" content="width=device-width, initial-scale=1.0">
  
  
  <link rel="shortcut icon" href="../img/favicon.ico">
  <title>模型库2 - PaddleSlim Docs</title>
  <link href='https://fonts.googleapis.com/css?family=Lato:400,700|Roboto+Slab:400,700|Inconsolata:400,700' rel='stylesheet' type='text/css'>

  <link rel="stylesheet" href="../css/theme.css" type="text/css" />
  <link rel="stylesheet" href="../css/theme_extra.css" type="text/css" />
  <link rel="stylesheet" href="//cdnjs.cloudflare.com/ajax/libs/highlight.js/9.12.0/styles/github.min.css">
  
  <script>
    // Current page data
    var mkdocs_page_name = "\u6a21\u578b\u5e932";
    var mkdocs_page_input_path = "model_zoo2.md";
    var mkdocs_page_url = null;
  </script>
  
  <script src="../js/jquery-2.1.1.min.js" defer></script>
  <script src="../js/modernizr-2.8.3.min.js" defer></script>
  <script src="//cdnjs.cloudflare.com/ajax/libs/highlight.js/9.12.0/highlight.min.js"></script>
  <script>hljs.initHighlightingOnLoad();</script> 
  
</head>

<body class="wy-body-for-nav" role="document">

  <div class="wy-grid-for-nav">

    
    <nav data-toggle="wy-nav-shift" class="wy-nav-side stickynav">
      <div class="wy-side-nav-search">
        <a href=".." class="icon icon-home"> PaddleSlim Docs</a>
        <div role="search">
  <form id ="rtd-search-form" class="wy-form" action="../search.html" method="get">
    <input type="text" name="q" placeholder="Search docs" title="Type search term here" />
  </form>
</div>
      </div>

      <div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
	<ul class="current">
	  
          
            <li class="toctree-l1">
		
    <a class="" href="..">Home</a>
	    </li>
          
            <li class="toctree-l1">
		
    <span class="caption-text">API</span>
    <ul class="subnav">
                <li class="">
                    
    <a class="" href="../api/quantization_api/">量化</a>
                </li>
                <li class="">
                    
    <a class="" href="../api/prune_api/">剪枝与敏感度</a>
                </li>
                <li class="">
                    
    <a class="" href="../api/analysis_api/">模型分析</a>
                </li>
                <li class="">
                    
    <a class="" href="../api/single_distiller_api/">知识蒸馏</a>
                </li>
                <li class="">
                    
    <a class="" href="../api/nas_api/">SA搜索</a>
                </li>
                <li class="">
                    
    <a class="" href="../api/search_space/">搜索空间</a>
                </li>
                <li class="">
                    
    <a class="" href="../table_latency/">硬件延时评估表</a>
                </li>
    </ul>
	    </li>
          
            <li class="toctree-l1">
		
    <span class="caption-text">教程</span>
    <ul class="subnav">
                <li class="">
                    
    <a class="" href="../tutorials/quant_post_demo/">离线量化</a>
                </li>
                <li class="">
                    
    <a class="" href="../tutorials/quant_aware_demo/">量化训练</a>
                </li>
                <li class="">
                    
    <a class="" href="../tutorials/quant_embedding_demo/">Embedding量化</a>
                </li>
                <li class="">
                    
    <a class="" href="../tutorials/nas_demo/">SA搜索</a>
                </li>
                <li class="">
                    
    <a class="" href="../tutorials/distillation_demo/">知识蒸馏</a>
                </li>
    </ul>
	    </li>
          
            <li class="toctree-l1">
		
    <a class="" href="../algo/algo/">算法原理</a>
	    </li>
          
            <li class="toctree-l1">
		
    <a class="" href="../model_zoo/">模型库</a>
	    </li>
          
            <li class="toctree-l1 current">
		
    <a class="current" href="./">模型库2</a>
    <ul class="subnav">
            
    <li class="toctree-l2"><a href="#1">1. 量化</a></li>
    
        <ul>
        
            <li><a class="toctree-l3" href="#11">1.1 图象分类</a></li>
        
139
            <li><a class="toctree-l3" href="#12">1.2 目标检测</a></li>
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
        
            <li><a class="toctree-l3" href="#13">1.3 图像分割</a></li>
        
        </ul>
    

    <li class="toctree-l2"><a href="#2">2. 剪枝</a></li>
    
        <ul>
        
            <li><a class="toctree-l3" href="#21">2.1 图像分类</a></li>
        
            <li><a class="toctree-l3" href="#22">2.2 目标检测</a></li>
        
            <li><a class="toctree-l3" href="#23">2.3 图像分割</a></li>
        
        </ul>
    

    <li class="toctree-l2"><a href="#3">3. 蒸馏</a></li>
    
        <ul>
        
            <li><a class="toctree-l3" href="#31">3.1 图象分类</a></li>
        
            <li><a class="toctree-l3" href="#32">3.2 目标检测</a></li>
        
        </ul>
    

    </ul>
	    </li>
          
        </ul>
      </div>
      &nbsp;
    </nav>

    <section data-toggle="wy-nav-shift" class="wy-nav-content-wrap">

      
      <nav class="wy-nav-top" role="navigation" aria-label="top navigation">
        <i data-toggle="wy-nav-top" class="fa fa-bars"></i>
        <a href="..">PaddleSlim Docs</a>
      </nav>

      
      <div class="wy-nav-content">
        <div class="rst-content">
          <div role="navigation" aria-label="breadcrumbs navigation">
  <ul class="wy-breadcrumbs">
    <li><a href="..">Docs</a> &raquo;</li>
    
      
    
    <li>模型库2</li>
    <li class="wy-breadcrumbs-aside">
      
        <a href="https://github.com/PaddlePaddle/PaddleSlim/edit/master/docs/model_zoo2.md"
          class="icon icon-github"> Edit on GitHub</a>
      
    </li>
  </ul>
  <hr/>
</div>
          <div role="main">
            <div class="section">
              
                <h2 id="1">1. 量化<a class="headerlink" href="#1" title="Permanent link">#</a></h2>
<h3 id="11">1.1 图象分类<a class="headerlink" href="#11" title="Permanent link">#</a></h3>
<p>数据集:ImageNet1000类</p>
<table>
<thead>
<tr>
<th align="center">Model</th>
215
<th align="center">压缩方法</th>
216 217
<th align="center">Top-1/Top-5</th>
<th align="center">模型大小(MB)</th>
218
<th align="center">下载</th>
219 220 221 222
</tr>
</thead>
<tbody>
<tr>
223 224
<td align="center">MobileNetV1</td>
<td align="center">-</td>
225
<td align="center">70.99%/89.68%</td>
226
<td align="center">xx</td>
227
<td align="center"><a href="">下载链接</a></td>
228 229
</tr>
<tr>
230 231
<td align="center">MobileNetV1</td>
<td align="center">quant_psot</td>
232 233 234
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
235 236
</tr>
<tr>
237 238
<td align="center">MobileNetV1</td>
<td align="center">quant_aware</td>
239 240 241
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
242 243
</tr>
<tr>
244 245
<td align="center">MobileNetV2</td>
<td align="center">-</td>
246
<td align="center">72.15%/90.65%</td>
247
<td align="center">xx</td>
248
<td align="center"><a href="">下载链接</a></td>
249 250
</tr>
<tr>
251 252
<td align="center">MobileNetV2</td>
<td align="center">quant_post</td>
253 254 255
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
256 257
</tr>
<tr>
258 259
<td align="center">MobileNetV2</td>
<td align="center">quant_aware</td>
260 261 262
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
263 264
</tr>
<tr>
265 266
<td align="center">ResNet50</td>
<td align="center">-</td>
267
<td align="center">76.50%/93.00%</td>
268
<td align="center">xx</td>
269
<td align="center"><a href="">下载链接</a></td>
270 271
</tr>
<tr>
272 273
<td align="center">ResNet50</td>
<td align="center">quant_post</td>
274 275 276
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
277 278
</tr>
<tr>
279 280
<td align="center">ResNet50</td>
<td align="center">quant_aware</td>
281 282 283
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
284 285 286
</tr>
</tbody>
</table>
287
<h3 id="12">1.2 目标检测<a class="headerlink" href="#12" title="Permanent link">#</a></h3>
288 289 290 291 292
<p>数据集:COCO 2017 </p>
<table>
<thead>
<tr>
<th align="center">Model</th>
293
<th align="center">压缩方法</th>
294
<th align="center">Image/GPU</th>
295 296 297
<th align="center">输入608 Box AP</th>
<th align="center">输入416 Box AP</th>
<th align="center">输入320 Box AP</th>
298
<th align="center">模型大小(MB)</th>
299
<th align="center">下载</th>
300 301 302 303
</tr>
</thead>
<tbody>
<tr>
304 305
<td align="center">MobileNet-V1-YOLOv3</td>
<td align="center">-</td>
306
<td align="center">8</td>
307 308 309
<td align="center">29.3</td>
<td align="center">29.3</td>
<td align="center">27.1</td>
310
<td align="center">xx</td>
311
<td align="center"><a href="">下载链接</a></td>
312 313
</tr>
<tr>
314 315
<td align="center">MobileNet-V1-YOLOv3</td>
<td align="center">quant_post</td>
316 317 318 319 320 321
<td align="center">8</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
322 323
</tr>
<tr>
324 325
<td align="center">MobileNet-V1-YOLOv3</td>
<td align="center">quant_aware</td>
326 327 328 329 330 331
<td align="center">8</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
332 333 334
</tr>
<tr>
<td align="center">R50-dcn-YOLOv3 FP32</td>
335
<td align="center">-</td>
336 337 338 339 340 341
<td align="center">8</td>
<td align="center">41.4</td>
<td align="center">-</td>
<td align="center">-</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
342 343
</tr>
<tr>
344 345
<td align="center">R50-dcn-YOLOv3</td>
<td align="center">quant_post</td>
346 347 348 349 350 351
<td align="center">8</td>
<td align="center">xx</td>
<td align="center">-</td>
<td align="center">-</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
352 353
</tr>
<tr>
354 355
<td align="center">R50-dcn-YOLOv3</td>
<td align="center">quant_aware</td>
356 357 358 359 360 361
<td align="center">8</td>
<td align="center">xx</td>
<td align="center">-</td>
<td align="center">-</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
362 363 364
</tr>
</tbody>
</table>
365
<p>数据集:WIDER-FACE</p>
366 367 368 369
<table>
<thead>
<tr>
<th align="center">Model</th>
370
<th align="center">压缩方法</th>
371
<th align="center">Image/GPU</th>
372 373 374 375
<th align="center">输入尺寸</th>
<th align="center">Easy/Medium/Hard</th>
<th align="center">模型大小(MB)</th>
<th align="center">下载</th>
376 377 378 379
</tr>
</thead>
<tbody>
<tr>
380 381
<td align="center">BlazeFace</td>
<td align="center">-</td>
382
<td align="center">8</td>
383 384
<td align="center">640</td>
<td align="center">0.915/0.892/0.797</td>
385
<td align="center">xx</td>
386
<td align="center"><a href="">下载链接</a></td>
387 388
</tr>
<tr>
389 390
<td align="center">BlazeFace</td>
<td align="center">quant_post</td>
391 392 393 394 395
<td align="center">8</td>
<td align="center">640</td>
<td align="center">xx/xx/xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
396 397
</tr>
<tr>
398 399
<td align="center">BlazeFace</td>
<td align="center">quant_aware</td>
400
<td align="center">8</td>
401 402 403 404
<td align="center">640</td>
<td align="center">xx/xx/xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
405 406
</tr>
<tr>
407 408
<td align="center">BlazeFace-Lite</td>
<td align="center">-</td>
409
<td align="center">8</td>
410
<td align="center">640</td>
411 412 413
<td align="center">0.909/0.885/0.781</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
414 415
</tr>
<tr>
416 417
<td align="center">BlazeFace-Lite</td>
<td align="center">quant_post</td>
418
<td align="center">8</td>
419
<td align="center">640</td>
420 421 422
<td align="center">xx/xx/xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
423 424
</tr>
<tr>
425 426
<td align="center">BlazeFace-Lite</td>
<td align="center">quant_aware</td>
427 428 429 430 431
<td align="center">8</td>
<td align="center">640</td>
<td align="center">xx/xx/xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
432 433
</tr>
<tr>
434 435
<td align="center">BlazeFace-NAS</td>
<td align="center">-</td>
436 437 438 439 440
<td align="center">8</td>
<td align="center">640</td>
<td align="center">0.837/0.807/0.658</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
441 442
</tr>
<tr>
443 444
<td align="center">BlazeFace-NAS</td>
<td align="center">quant_post</td>
445 446 447 448 449
<td align="center">8</td>
<td align="center">640</td>
<td align="center">xx/xx/xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
450 451
</tr>
<tr>
452 453
<td align="center">BlazeFace-NAS</td>
<td align="center">quant_aware</td>
454 455 456 457 458
<td align="center">8</td>
<td align="center">640</td>
<td align="center">xx/xx/xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
459 460 461 462 463 464 465 466 467
</tr>
</tbody>
</table>
<h3 id="13">1.3 图像分割<a class="headerlink" href="#13" title="Permanent link">#</a></h3>
<p>数据集:Cityscapes</p>
<table>
<thead>
<tr>
<th align="center">Model</th>
468
<th align="center">压缩方法</th>
469 470 471
<th align="center">mIoU</th>
<th align="center">模型大小(MB)</th>
<th align="center">下载</th>
472 473 474 475 476
</tr>
</thead>
<tbody>
<tr>
<td align="center">DeepLabv3+/MobileNetv1</td>
477
<td align="center">-</td>
478 479 480
<td align="center">63.26</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
481 482
</tr>
<tr>
483 484
<td align="center">DeepLabv3+/MobileNetv1</td>
<td align="center">quant_post</td>
485 486 487
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
488 489
</tr>
<tr>
490 491
<td align="center">DeepLabv3+/MobileNetv1</td>
<td align="center">quant_aware</td>
492 493 494
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
495 496
</tr>
<tr>
497
<td align="center">DeepLabv3+/MobileNetv2</td>
498
<td align="center">-</td>
499 500 501
<td align="center">69.81</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
502 503
</tr>
<tr>
504 505
<td align="center">DeepLabv3+/MobileNetv2</td>
<td align="center">quant_post</td>
506 507 508 509 510
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
</tr>
<tr>
511 512
<td align="center">DeepLabv3+/MobileNetv2</td>
<td align="center">quant_aware</td>
513 514 515
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
516 517 518 519 520 521 522 523 524 525
</tr>
</tbody>
</table>
<h2 id="2">2. 剪枝<a class="headerlink" href="#2" title="Permanent link">#</a></h2>
<h3 id="21">2.1 图像分类<a class="headerlink" href="#21" title="Permanent link">#</a></h3>
<p>数据集:ImageNet1000类</p>
<table>
<thead>
<tr>
<th align="center">Model</th>
526
<th align="center">压缩方法</th>
527
<th align="center">Top-1/Top-5</th>
528 529 530
<th align="center">模型大小(MB)</th>
<th align="center">FLOPs</th>
<th align="center">下载</th>
531 532 533 534 535
</tr>
</thead>
<tbody>
<tr>
<td align="center">MobileNetV1</td>
536
<td align="center">-</td>
537 538 539 540
<td align="center">70.99%/89.68%</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
541 542
</tr>
<tr>
543 544
<td align="center">MobileNetV1</td>
<td align="center">uniform -xx%</td>
545 546 547 548
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
549 550
</tr>
<tr>
551 552
<td align="center">MobileNetV1</td>
<td align="center">sensitive -xx%</td>
553 554 555 556
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
557 558 559
</tr>
<tr>
<td align="center">MobileNetV2</td>
560
<td align="center">-</td>
561 562 563 564
<td align="center">72.15%/90.65%</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
565 566
</tr>
<tr>
567 568
<td align="center">MobileNetV2</td>
<td align="center">uniform -xx%</td>
569 570 571 572
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
573 574
</tr>
<tr>
575 576
<td align="center">MobileNetV2</td>
<td align="center">sensitive -xx%</td>
577 578 579 580
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
581 582 583
</tr>
<tr>
<td align="center">ResNet34</td>
584
<td align="center">-</td>
585 586 587 588
<td align="center">74.57%/92.14%</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
589 590
</tr>
<tr>
591 592
<td align="center">ResNet34</td>
<td align="center">uniform -xx%</td>
593
<td align="center">xx%/xx%</td>
594 595
<td align="center">xx</td>
<td align="center">xx</td>
596
<td align="center"><a href="">下载链接</a></td>
597 598
</tr>
<tr>
599 600
<td align="center">ResNet34</td>
<td align="center">auto -xx%</td>
601
<td align="center">xx%/xx%</td>
602 603
<td align="center">xx</td>
<td align="center">xx</td>
604
<td align="center"><a href="">下载链接</a></td>
605 606 607 608 609 610 611 612 613
</tr>
</tbody>
</table>
<h3 id="22">2.2 目标检测<a class="headerlink" href="#22" title="Permanent link">#</a></h3>
<p>数据集:Pasacl VOC &amp; COCO 2017</p>
<table>
<thead>
<tr>
<th align="center">Model</th>
614
<th>压缩方法</th>
615 616
<th align="center">数据集</th>
<th align="center">Image/GPU</th>
617 618 619 620 621 622
<th align="center">输入608 mAP</th>
<th align="center">输入416 mAP</th>
<th align="center">输入320  mAP</th>
<th align="center">模型大小(MB)</th>
<th align="center">FLOPs</th>
<th align="center">下载</th>
623 624 625 626 627
</tr>
</thead>
<tbody>
<tr>
<td align="center">MobileNet-V1-YOLOv3</td>
628
<td>-</td>
629 630
<td align="center">Pasacl VOC</td>
<td align="center">8</td>
631 632 633 634 635 636
<td align="center">76.2</td>
<td align="center">76.7</td>
<td align="center">75.3</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
637 638
</tr>
<tr>
639 640
<td align="center">MobileNet-V1-YOLOv3</td>
<td>uniform  -xx%</td>
641
<td align="center">Pasacl VOC</td>
642 643 644 645 646 647 648
<td align="center">8</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
649 650 651
</tr>
<tr>
<td align="center">MobileNet-V1-YOLOv3</td>
652
<td>-</td>
653
<td align="center">COCO</td>
654 655 656 657 658 659 660
<td align="center">8</td>
<td align="center">29.3</td>
<td align="center">29.3</td>
<td align="center">27.1</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
661 662
</tr>
<tr>
663 664
<td align="center">MobileNet-V1-YOLOv3</td>
<td>uniform -xx%</td>
665
<td align="center">COCO</td>
666 667 668 669 670 671 672
<td align="center">8</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
673 674 675
</tr>
<tr>
<td align="center">R50-dcn-YOLOv3</td>
676
<td>-</td>
677
<td align="center">COCO</td>
678 679 680 681
<td align="center">8</td>
<td align="center">41.4</td>
<td align="center">-</td>
<td align="center">-</td>
682 683
<td align="center">xx</td>
<td align="center">xx</td>
684
<td align="center"><a href="">下载链接</a></td>
685 686
</tr>
<tr>
687 688
<td align="center">R50-dcn-YOLOv3</td>
<td>uniform -xx%</td>
689 690 691 692 693
<td align="center">COCO</td>
<td align="center">8</td>
<td align="center">xx</td>
<td align="center">-</td>
<td align="center">-</td>
694 695
<td align="center">xx</td>
<td align="center">xx</td>
696
<td align="center"><a href="">下载链接</a></td>
697 698 699 700 701 702 703 704 705
</tr>
</tbody>
</table>
<h3 id="23">2.3 图像分割<a class="headerlink" href="#23" title="Permanent link">#</a></h3>
<p>数据集:Cityscapes</p>
<table>
<thead>
<tr>
<th align="center">Model</th>
706
<th align="center">压缩方法</th>
707 708 709 710
<th align="center">mIoU</th>
<th align="center">模型大小(MB)</th>
<th align="center">FLOPs</th>
<th align="center">下载</th>
711 712 713 714 715
</tr>
</thead>
<tbody>
<tr>
<td align="center">DeepLabv3+/MobileNetv2</td>
716
<td align="center">-</td>
717 718 719 720
<td align="center">69.81</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
721 722
</tr>
<tr>
723 724
<td align="center">DeepLabv3+/MobileNetv2</td>
<td align="center">prune -xx%</td>
725
<td align="center">xx</td>
726 727
<td align="center">xx</td>
<td align="center">xx</td>
728
<td align="center"><a href="">下载链接</a></td>
729 730 731 732 733 734 735 736 737 738
</tr>
</tbody>
</table>
<h2 id="3">3. 蒸馏<a class="headerlink" href="#3" title="Permanent link">#</a></h2>
<h3 id="31">3.1 图象分类<a class="headerlink" href="#31" title="Permanent link">#</a></h3>
<p>数据集:ImageNet1000类</p>
<table>
<thead>
<tr>
<th align="center">Model</th>
739
<th align="center">蒸馏 teacher</th>
740
<th align="center">baseline</th>
741
<th align="center">下载</th>
742 743 744 745 746
</tr>
</thead>
<tbody>
<tr>
<td align="center">MobileNetV1</td>
747
<td align="center">-</td>
748 749 750 751
<td align="center">70.99%/89.68%</td>
<td align="center"><a href="http://paddle-imagenet-models-name.bj.bcebos.com/MobileNetV1_pretrained.tar">下载链接</a></td>
</tr>
<tr>
752 753
<td align="center">MobileNetV1</td>
<td align="center">ResNet50_vd<sup><a href="#trans1">1</a></sup></td>
754 755
<td align="center">72.79%/90.69%</td>
<td align="center"><a href="">下载链接</a></td>
756 757 758
</tr>
<tr>
<td align="center">MobileNetV2</td>
759
<td align="center">-</td>
760 761 762 763
<td align="center">72.15%/90.65%</td>
<td align="center"><a href="">下载链接</a></td>
</tr>
<tr>
764 765
<td align="center">MobileNetV2</td>
<td align="center">ResNet50_vd<sup><a href="#trans1">1</a></sup></td>
766 767
<td align="center">74.30%/91.52%</td>
<td align="center"><a href="">下载链接</a></td>
768 769 770
</tr>
<tr>
<td align="center">ResNet50</td>
771
<td align="center">-</td>
772 773 774 775
<td align="center">76.50%/93.00%</td>
<td align="center"><a href="">下载链接</a></td>
</tr>
<tr>
776 777
<td align="center">ResNet50</td>
<td align="center">ResNet101<sup><a href="#trans2">2</a></sup></td>
778 779
<td align="center">77.40%/93.48%</td>
<td align="center"><a href="">下载链接</a></td>
780 781 782 783 784 785 786 787 788 789 790 791 792 793 794
</tr>
</tbody>
</table>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p><a name="trans1">  [1]</a><a href="https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_pretrained.tar">ResNet50_vd</a>预训练模型Top-1/Top-5准确率分别为79.12%/94.44%</p>
<p>带_vd后缀代表开启了Mixup训练,Mixup相关介绍参考<a href="https://arxiv.org/abs/1710.09412">mixup: Beyond Empirical Risk Minimization</a></p>
<p><a name="trans1">[2]</a><a href="https://paddle-imagenet-models-name.bj.bcebos.com/ResNet101_pretrained.tar">ResNet101</a>预训练模型Top-1/Top-5准确率分别为77.56%/93.64%</p>
</div>
<h3 id="32">3.2 目标检测<a class="headerlink" href="#32" title="Permanent link">#</a></h3>
<p>数据集:Pasacl VOC &amp; COCO 2017</p>
<table>
<thead>
<tr>
<th align="center">Model</th>
795
<th align="center">蒸馏 teacher</th>
796 797
<th align="center">数据集</th>
<th align="center">Image/GPU</th>
798 799 800 801
<th align="center">输入640 mAP</th>
<th align="center">输入416 mAP</th>
<th align="center">输入320 mAP</th>
<th align="center">下载链接</th>
802 803 804 805 806
</tr>
</thead>
<tbody>
<tr>
<td align="center">MobileNet-V1-YOLOv3</td>
807
<td align="center">-</td>
808 809
<td align="center">Pasacl VOC</td>
<td align="center">16</td>
810 811 812 813
<td align="center">76.2</td>
<td align="center">76.7</td>
<td align="center">75.3</td>
<td align="center"><a href="">下载链接</a></td>
814 815
</tr>
<tr>
816 817
<td align="center">MobileNet-V1-YOLOv3</td>
<td align="center">ResNet34-YOLOv3-VOC<sup><a href="#trans3">3</a></sup></td>
818
<td align="center">Pasacl VOC</td>
819 820 821 822 823
<td align="center">16</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
824 825 826
</tr>
<tr>
<td align="center">MobileNet-V1-YOLOv3</td>
827
<td align="center">-</td>
828
<td align="center">COCO</td>
829 830 831 832 833
<td align="center">16</td>
<td align="center">29.3</td>
<td align="center">29.3</td>
<td align="center">27.1</td>
<td align="center"><a href="">下载链接</a></td>
834 835
</tr>
<tr>
836 837
<td align="center">MobileNet-V1-YOLOv3</td>
<td align="center">ResNet34-YOLOv3-COCO<sup><a href="#trans4">4</a></sup></td>
838
<td align="center">COCO</td>
839 840 841 842 843
<td align="center">16</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900
</tr>
</tbody>
</table>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p><a name="trans1">[3]</a><a href="">ResNet34-YOLOv3-VOC</a>预训练模型的Box AP为82.6</p>
<p><a name="trans1">[4]</a><a href="">ResNet34-YOLOv3-COCO</a>预训练模型的Box AP为36.2</p>
</div>
              
            </div>
          </div>
          <footer>
  
    <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
      
      
        <a href="../model_zoo/" class="btn btn-neutral" title="模型库"><span class="icon icon-circle-arrow-left"></span> Previous</a>
      
    </div>
  

  <hr/>

  <div role="contentinfo">
    <!-- Copyright etc -->
    
  </div>

  Built with <a href="http://www.mkdocs.org">MkDocs</a> using a <a href="https://github.com/snide/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
</footer>
      
        </div>
      </div>

    </section>

  </div>

  <div class="rst-versions" role="note" style="cursor: pointer">
    <span class="rst-current-version" data-toggle="rst-current-version">
      
          <a href="https://github.com/PaddlePaddle/PaddleSlim/" class="fa fa-github" style="float: left; color: #fcfcfc"> GitHub</a>
      
      
        <span><a href="../model_zoo/" style="color: #fcfcfc;">&laquo; Previous</a></span>
      
      
    </span>
</div>
    <script>var base_url = '..';</script>
    <script src="../js/theme.js" defer></script>
      <script src="../mathjax-config.js" defer></script>
      <script src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.0/MathJax.js?config=TeX-AMS-MML_HTMLorMML" defer></script>
      <script src="../search/main.js" defer></script>

</body>
</html>