index.html 25.8 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
        
            <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>
          
173 174 175 176 177
            <li class="toctree-l1">
		
    <a class="" href="../model_zoo3/">模型库3</a>
	    </li>
          
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
        </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>
220
<th align="center">压缩方法</th>
221 222
<th align="center">Top-1/Top-5</th>
<th align="center">模型大小(MB)</th>
223
<th align="center">下载</th>
224 225 226 227
</tr>
</thead>
<tbody>
<tr>
228 229
<td align="center">MobileNetV1</td>
<td align="center">-</td>
230
<td align="center">70.99%/89.68%</td>
231
<td align="center">xx</td>
232
<td align="center"><a href="">下载链接</a></td>
233 234
</tr>
<tr>
235 236
<td align="center">MobileNetV1</td>
<td align="center">quant_psot</td>
237 238 239
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
240 241
</tr>
<tr>
242 243
<td align="center">MobileNetV1</td>
<td align="center">quant_aware</td>
244 245 246
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
247 248
</tr>
<tr>
249 250
<td align="center">MobileNetV2</td>
<td align="center">-</td>
251
<td align="center">72.15%/90.65%</td>
252
<td align="center">xx</td>
253
<td align="center"><a href="">下载链接</a></td>
254 255
</tr>
<tr>
256 257
<td align="center">MobileNetV2</td>
<td align="center">quant_post</td>
258 259 260
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
261 262
</tr>
<tr>
263 264
<td align="center">MobileNetV2</td>
<td align="center">quant_aware</td>
265 266 267
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
268 269
</tr>
<tr>
270 271
<td align="center">ResNet50</td>
<td align="center">-</td>
272
<td align="center">76.50%/93.00%</td>
273
<td align="center">xx</td>
274
<td align="center"><a href="">下载链接</a></td>
275 276
</tr>
<tr>
277 278
<td align="center">ResNet50</td>
<td align="center">quant_post</td>
279 280 281
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
282 283
</tr>
<tr>
284 285
<td align="center">ResNet50</td>
<td align="center">quant_aware</td>
286 287 288
<td align="center">xx%/xx%</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
289 290 291
</tr>
</tbody>
</table>
292
<h3 id="12">1.2 目标检测<a class="headerlink" href="#12" title="Permanent link">#</a></h3>
293 294 295 296 297
<p>数据集:COCO 2017 </p>
<table>
<thead>
<tr>
<th align="center">Model</th>
298
<th align="center">压缩方法</th>
299
<th align="center">Image/GPU</th>
300 301 302
<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>
303
<th align="center">模型大小(MB)</th>
304
<th align="center">下载</th>
305 306 307 308
</tr>
</thead>
<tbody>
<tr>
309 310
<td align="center">MobileNet-V1-YOLOv3</td>
<td align="center">-</td>
311
<td align="center">8</td>
312 313 314
<td align="center">29.3</td>
<td align="center">29.3</td>
<td align="center">27.1</td>
315
<td align="center">xx</td>
316
<td align="center"><a href="">下载链接</a></td>
317 318
</tr>
<tr>
319 320
<td align="center">MobileNet-V1-YOLOv3</td>
<td align="center">quant_post</td>
321 322 323 324 325 326
<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>
327 328
</tr>
<tr>
329 330
<td align="center">MobileNet-V1-YOLOv3</td>
<td align="center">quant_aware</td>
331 332 333 334 335 336
<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>
337 338 339
</tr>
<tr>
<td align="center">R50-dcn-YOLOv3 FP32</td>
340
<td align="center">-</td>
341 342
<td align="center">8</td>
<td align="center">41.4</td>
343 344
<td align="center">xx</td>
<td align="center">xx</td>
345 346
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
347 348
</tr>
<tr>
349 350
<td align="center">R50-dcn-YOLOv3</td>
<td align="center">quant_post</td>
351 352
<td align="center">8</td>
<td align="center">xx</td>
353 354
<td align="center">xx</td>
<td align="center">xx</td>
355 356
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
357 358
</tr>
<tr>
359 360
<td align="center">R50-dcn-YOLOv3</td>
<td align="center">quant_aware</td>
361 362
<td align="center">8</td>
<td align="center">xx</td>
363 364
<td align="center">xx</td>
<td align="center">xx</td>
365 366
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
367 368 369
</tr>
</tbody>
</table>
370
<p>数据集:WIDER-FACE</p>
371 372 373 374
<table>
<thead>
<tr>
<th align="center">Model</th>
375
<th align="center">压缩方法</th>
376
<th align="center">Image/GPU</th>
377 378 379 380
<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>
381 382 383 384
</tr>
</thead>
<tbody>
<tr>
385 386
<td align="center">BlazeFace</td>
<td align="center">-</td>
387
<td align="center">8</td>
388 389
<td align="center">640</td>
<td align="center">0.915/0.892/0.797</td>
390
<td align="center">xx</td>
391
<td align="center"><a href="">下载链接</a></td>
392 393
</tr>
<tr>
394 395
<td align="center">BlazeFace</td>
<td align="center">quant_post</td>
396 397 398 399 400
<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>
401 402
</tr>
<tr>
403 404
<td align="center">BlazeFace</td>
<td align="center">quant_aware</td>
405
<td align="center">8</td>
406 407 408 409
<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>
410 411
</tr>
<tr>
412 413
<td align="center">BlazeFace-Lite</td>
<td align="center">-</td>
414
<td align="center">8</td>
415
<td align="center">640</td>
416 417 418
<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>
419 420
</tr>
<tr>
421 422
<td align="center">BlazeFace-Lite</td>
<td align="center">quant_post</td>
423
<td align="center">8</td>
424
<td align="center">640</td>
425 426 427
<td align="center">xx/xx/xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
428 429
</tr>
<tr>
430 431
<td align="center">BlazeFace-Lite</td>
<td align="center">quant_aware</td>
432 433 434 435 436
<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>
437 438
</tr>
<tr>
439 440
<td align="center">BlazeFace-NAS</td>
<td align="center">-</td>
441 442 443 444 445
<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>
446 447
</tr>
<tr>
448 449
<td align="center">BlazeFace-NAS</td>
<td align="center">quant_post</td>
450 451 452 453 454
<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>
455 456
</tr>
<tr>
457 458
<td align="center">BlazeFace-NAS</td>
<td align="center">quant_aware</td>
459 460 461 462 463
<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>
464 465 466 467 468 469 470 471 472
</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>
473
<th align="center">压缩方法</th>
474 475 476
<th align="center">mIoU</th>
<th align="center">模型大小(MB)</th>
<th align="center">下载</th>
477 478 479 480 481
</tr>
</thead>
<tbody>
<tr>
<td align="center">DeepLabv3+/MobileNetv1</td>
482
<td align="center">-</td>
483 484 485
<td align="center">63.26</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
486 487
</tr>
<tr>
488 489
<td align="center">DeepLabv3+/MobileNetv1</td>
<td align="center">quant_post</td>
490 491 492
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
493 494
</tr>
<tr>
495 496
<td align="center">DeepLabv3+/MobileNetv1</td>
<td align="center">quant_aware</td>
497 498 499
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
500 501
</tr>
<tr>
502
<td align="center">DeepLabv3+/MobileNetv2</td>
503
<td align="center">-</td>
504 505 506
<td align="center">69.81</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
507 508
</tr>
<tr>
509 510
<td align="center">DeepLabv3+/MobileNetv2</td>
<td align="center">quant_post</td>
511 512 513 514 515
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
</tr>
<tr>
516 517
<td align="center">DeepLabv3+/MobileNetv2</td>
<td align="center">quant_aware</td>
518 519 520
<td align="center">xx</td>
<td align="center">xx</td>
<td align="center"><a href="">下载链接</a></td>
521 522 523 524 525 526 527 528 529 530
</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>
531
<th align="center">压缩方法</th>
532
<th align="center">Top-1/Top-5</th>
533 534 535
<th align="center">模型大小(MB)</th>
<th align="center">FLOPs</th>
<th align="center">下载</th>
536 537 538 539 540
</tr>
</thead>
<tbody>
<tr>
<td align="center">MobileNetV1</td>
541
<td align="center">-</td>
542 543 544 545
<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>
546 547
</tr>
<tr>
548 549
<td align="center">MobileNetV1</td>
<td align="center">uniform -xx%</td>
550 551 552 553
<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>
554 555
</tr>
<tr>
556 557
<td align="center">MobileNetV1</td>
<td align="center">sensitive -xx%</td>
558 559 560 561
<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>
562 563 564
</tr>
<tr>
<td align="center">MobileNetV2</td>
565
<td align="center">-</td>
566 567 568 569
<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>
570 571
</tr>
<tr>
572 573
<td align="center">MobileNetV2</td>
<td align="center">uniform -xx%</td>
574 575 576 577
<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>
578 579
</tr>
<tr>
580 581
<td align="center">MobileNetV2</td>
<td align="center">sensitive -xx%</td>
582 583 584 585
<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>
586 587 588
</tr>
<tr>
<td align="center">ResNet34</td>
589
<td align="center">-</td>
590 591 592 593
<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>
594 595
</tr>
<tr>
596 597
<td align="center">ResNet34</td>
<td align="center">uniform -xx%</td>
598
<td align="center">xx%/xx%</td>
599 600
<td align="center">xx</td>
<td align="center">xx</td>
601
<td align="center"><a href="">下载链接</a></td>
602 603
</tr>
<tr>
604 605
<td align="center">ResNet34</td>
<td align="center">auto -xx%</td>
606
<td align="center">xx%/xx%</td>
607 608
<td align="center">xx</td>
<td align="center">xx</td>
609
<td align="center"><a href="">下载链接</a></td>
610 611 612 613 614 615 616 617 618
</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>
619
<th>压缩方法</th>
620 621
<th align="center">数据集</th>
<th align="center">Image/GPU</th>
622 623 624 625 626 627
<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>
628 629 630 631 632
</tr>
</thead>
<tbody>
<tr>
<td align="center">MobileNet-V1-YOLOv3</td>
633
<td>-</td>
634 635
<td align="center">Pasacl VOC</td>
<td align="center">8</td>
636 637 638 639 640 641
<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>
642 643
</tr>
<tr>
644 645
<td align="center">MobileNet-V1-YOLOv3</td>
<td>uniform  -xx%</td>
646
<td align="center">Pasacl VOC</td>
647 648 649 650 651 652 653
<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>
654 655 656
</tr>
<tr>
<td align="center">MobileNet-V1-YOLOv3</td>
657
<td>-</td>
658
<td align="center">COCO</td>
659 660 661 662 663 664 665
<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>
666 667
</tr>
<tr>
668 669
<td align="center">MobileNet-V1-YOLOv3</td>
<td>uniform -xx%</td>
670
<td align="center">COCO</td>
671 672 673 674 675 676 677
<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>
678 679 680
</tr>
<tr>
<td align="center">R50-dcn-YOLOv3</td>
681
<td>-</td>
682
<td align="center">COCO</td>
683 684
<td align="center">8</td>
<td align="center">41.4</td>
685 686
<td align="center">xx</td>
<td align="center">xx</td>
687 688
<td align="center">xx</td>
<td align="center">xx</td>
689
<td align="center"><a href="">下载链接</a></td>
690 691
</tr>
<tr>
692 693
<td align="center">R50-dcn-YOLOv3</td>
<td>uniform -xx%</td>
694 695 696
<td align="center">COCO</td>
<td align="center">8</td>
<td align="center">xx</td>
697 698
<td align="center">xx</td>
<td align="center">xx</td>
699 700
<td align="center">xx</td>
<td align="center">xx</td>
701
<td align="center"><a href="">下载链接</a></td>
702 703 704 705 706 707 708 709 710
</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>
711
<th align="center">压缩方法</th>
712 713 714 715
<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>
716 717 718 719 720
</tr>
</thead>
<tbody>
<tr>
<td align="center">DeepLabv3+/MobileNetv2</td>
721
<td align="center">-</td>
722 723 724 725
<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>
726 727
</tr>
<tr>
728 729
<td align="center">DeepLabv3+/MobileNetv2</td>
<td align="center">prune -xx%</td>
730
<td align="center">xx</td>
731 732
<td align="center">xx</td>
<td align="center">xx</td>
733
<td align="center"><a href="">下载链接</a></td>
734 735 736 737 738 739 740 741 742 743
</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>
744
<th align="center">蒸馏 teacher</th>
745
<th align="center">Top-1/Top-5</th>
746
<th align="center">下载</th>
747 748 749 750 751
</tr>
</thead>
<tbody>
<tr>
<td align="center">MobileNetV1</td>
752
<td align="center">-</td>
753 754 755 756
<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>
757 758
<td align="center">MobileNetV1</td>
<td align="center">ResNet50_vd<sup><a href="#trans1">1</a></sup></td>
759 760
<td align="center">72.79%/90.69%</td>
<td align="center"><a href="">下载链接</a></td>
761 762 763
</tr>
<tr>
<td align="center">MobileNetV2</td>
764
<td align="center">-</td>
765 766 767 768
<td align="center">72.15%/90.65%</td>
<td align="center"><a href="">下载链接</a></td>
</tr>
<tr>
769 770
<td align="center">MobileNetV2</td>
<td align="center">ResNet50_vd<sup><a href="#trans1">1</a></sup></td>
771 772
<td align="center">74.30%/91.52%</td>
<td align="center"><a href="">下载链接</a></td>
773 774 775
</tr>
<tr>
<td align="center">ResNet50</td>
776
<td align="center">-</td>
777 778 779 780
<td align="center">76.50%/93.00%</td>
<td align="center"><a href="">下载链接</a></td>
</tr>
<tr>
781 782
<td align="center">ResNet50</td>
<td align="center">ResNet101<sup><a href="#trans2">2</a></sup></td>
783 784
<td align="center">77.40%/93.48%</td>
<td align="center"><a href="">下载链接</a></td>
785 786 787 788 789 790 791 792 793 794 795 796 797 798 799
</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>
800
<th align="center">蒸馏 teacher</th>
801 802
<th align="center">数据集</th>
<th align="center">Image/GPU</th>
803 804 805 806
<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>
807 808 809 810 811
</tr>
</thead>
<tbody>
<tr>
<td align="center">MobileNet-V1-YOLOv3</td>
812
<td align="center">-</td>
813 814
<td align="center">Pasacl VOC</td>
<td align="center">16</td>
815 816 817 818
<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>
819 820
</tr>
<tr>
821 822
<td align="center">MobileNet-V1-YOLOv3</td>
<td align="center">ResNet34-YOLOv3-VOC<sup><a href="#trans3">3</a></sup></td>
823
<td align="center">Pasacl VOC</td>
824 825 826 827 828
<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>
829 830 831
</tr>
<tr>
<td align="center">MobileNet-V1-YOLOv3</td>
832
<td align="center">-</td>
833
<td align="center">COCO</td>
834 835 836 837 838
<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>
839 840
</tr>
<tr>
841 842
<td align="center">MobileNet-V1-YOLOv3</td>
<td align="center">ResNet34-YOLOv3-COCO<sup><a href="#trans4">4</a></sup></td>
843
<td align="center">COCO</td>
844 845 846 847 848
<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>
849 850 851 852 853 854 855 856 857 858 859 860 861 862 863
</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">
      
864 865
        <a href="../model_zoo3/" class="btn btn-neutral float-right" title="模型库3">Next <span class="icon icon-circle-arrow-right"></span></a>
      
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
      
        <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>
      
      
898 899
        <span style="margin-left: 15px"><a href="../model_zoo3/" style="color: #fcfcfc">Next &raquo;</a></span>
      
900 901 902 903 904 905 906 907 908 909
    </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>