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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173
|
<!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="canonical" href="http://keras.io/applications/">
<link rel="shortcut icon" href="../img/favicon.ico">
<title>Applications - Keras Documentation</title>
<link href='https://fonts.googleapis.com/css?family=Lato:400,700|Source+Sans+Pro: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="https://cdnjs.cloudflare.com/ajax/libs/highlight.js/9.12.0/styles/github.min.css">
<script>
// Current page data
var mkdocs_page_name = "Applications";
var mkdocs_page_input_path = "applications.md";
var mkdocs_page_url = "/applications/";
</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="https://cdnjs.cloudflare.com/ajax/libs/highlight.js/9.12.0/highlight.min.js"></script>
<script>hljs.initHighlightingOnLoad();</script>
<script>
(function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){
(i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o),
m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)
})(window,document,'script','https://www.google-analytics.com/analytics.js','ga');
ga('create', 'UA-61785484-1', 'keras.io');
ga('send', 'pageview');
</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-scroll">
<a href="">
<div class="keras-logo">
<img src="/img/keras-logo-small.jpg" class="keras-logo-img">
Keras Documentation
</div>
</a>
<div class="wy-side-nav-search">
<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>
<li class="toctree-l1"><a class="reference internal" href="..">Home</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../why-use-keras/">Why use Keras</a>
</li>
</ul>
<p class="caption"><span class="caption-text">Getting started</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../getting-started/sequential-model-guide/">Guide to the Sequential model</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../getting-started/functional-api-guide/">Guide to the Functional API</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../getting-started/faq/">FAQ</a>
</li>
</ul>
<p class="caption"><span class="caption-text">Models</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../models/about-keras-models/">About Keras models</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../models/sequential/">Sequential</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../models/model/">Model (functional API)</a>
</li>
</ul>
<p class="caption"><span class="caption-text">Layers</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../layers/about-keras-layers/">About Keras layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/core/">Core Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/convolutional/">Convolutional Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/pooling/">Pooling Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/local/">Locally-connected Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/recurrent/">Recurrent Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/embeddings/">Embedding Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/merge/">Merge Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/advanced-activations/">Advanced Activations Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/normalization/">Normalization Layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/noise/">Noise layers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/wrappers/">Layer wrappers</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../layers/writing-your-own-keras-layers/">Writing your own Keras layers</a>
</li>
</ul>
<p class="caption"><span class="caption-text">Preprocessing</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../preprocessing/sequence/">Sequence Preprocessing</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../preprocessing/text/">Text Preprocessing</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../preprocessing/image/">Image Preprocessing</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../losses/">Losses</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../metrics/">Metrics</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../optimizers/">Optimizers</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../activations/">Activations</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../callbacks/">Callbacks</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../datasets/">Datasets</a>
</li>
</ul>
<ul class="current">
<li class="toctree-l1 current"><a class="reference internal current" href="./">Applications</a>
<ul class="current">
<li class="toctree-l2"><a class="reference internal" href="#available-models">Available models</a>
<ul>
<li class="toctree-l3"><a class="reference internal" href="#models-for-image-classification-with-weights-trained-on-imagenet">Models for image classification with weights trained on ImageNet:</a>
</li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="#usage-examples-for-image-classification-models">Usage examples for image classification models</a>
<ul>
<li class="toctree-l3"><a class="reference internal" href="#classify-imagenet-classes-with-resnet50">Classify ImageNet classes with ResNet50</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#extract-features-with-vgg16">Extract features with VGG16</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#extract-features-from-an-arbitrary-intermediate-layer-with-vgg19">Extract features from an arbitrary intermediate layer with VGG19</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#fine-tune-inceptionv3-on-a-new-set-of-classes">Fine-tune InceptionV3 on a new set of classes</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#build-inceptionv3-over-a-custom-input-tensor">Build InceptionV3 over a custom input tensor</a>
</li>
</ul>
</li>
</ul>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../backend/">Backend</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../initializers/">Initializers</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../regularizers/">Regularizers</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../constraints/">Constraints</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../visualization/">Visualization</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../scikit-learn-api/">Scikit-learn API</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../utils/">Utils</a>
</li>
</ul>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../contributing/">Contributing</a>
</li>
</ul>
<p class="caption"><span class="caption-text">Examples</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../examples/addition_rnn/">Addition RNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/antirectifier/">Custom layer - antirectifier</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/babi_rnn/">Baby RNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/babi_memnn/">Baby MemNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/cifar10_cnn/">CIFAR-10 CNN</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/cifar10_resnet/">CIFAR-10 ResNet</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/conv_filter_visualization/">Convolution filter visualization</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/conv_lstm/">Convolutional LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/deep_dream/">Deep Dream</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/image_ocr/">Image OCR</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/imdb_bidirectional_lstm/">Bidirectional LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/imdb_cnn/">1D CNN for text classification</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/imdb_cnn_lstm/">Sentiment classification CNN-LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/imdb_fasttext/">Fasttext for text classification</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/imdb_lstm/">Sentiment classification LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/lstm_seq2seq/">Sequence to sequence - training</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/lstm_seq2seq_restore/">Sequence to sequence - prediction</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/lstm_stateful/">Stateful LSTM</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/lstm_text_generation/">LSTM for text generation</a>
</li>
<li class="toctree-l1"><a class="reference internal" href="../examples/mnist_acgan/">Auxiliary Classifier GAN</a>
</li>
</ul>
</div>
</div>
</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="..">Keras Documentation</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> »</li>
<li>Applications</li>
<li class="wy-breadcrumbs-aside">
<a href="https://github.com/keras-team/keras/tree/master/docs"
class="icon icon-github"> Edit on GitHub</a>
</li>
</ul>
<hr/>
</div>
<div role="main">
<div class="section">
<h1 id="applications">Applications</h1>
<p>Keras Applications are deep learning models that are made available alongside pre-trained weights.
These models can be used for prediction, feature extraction, and fine-tuning.</p>
<p>Weights are downloaded automatically when instantiating a model. They are stored at <code>~/.keras/models/</code>.</p>
<h2 id="available-models">Available models</h2>
<h3 id="models-for-image-classification-with-weights-trained-on-imagenet">Models for image classification with weights trained on ImageNet:</h3>
<ul>
<li><a href="#xception">Xception</a></li>
<li><a href="#vgg16">VGG16</a></li>
<li><a href="#vgg19">VGG19</a></li>
<li><a href="#resnet">ResNet, ResNetV2</a></li>
<li><a href="#inceptionv3">InceptionV3</a></li>
<li><a href="#inceptionresnetv2">InceptionResNetV2</a></li>
<li><a href="#mobilenet">MobileNet</a></li>
<li><a href="#mobilenetv2">MobileNetV2</a></li>
<li><a href="#densenet">DenseNet</a></li>
<li><a href="#nasnet">NASNet</a></li>
</ul>
<p>All of these architectures are compatible with all the backends (TensorFlow, Theano, and CNTK), and upon instantiation the models will be built according to the image data format set in your Keras configuration file at <code>~/.keras/keras.json</code>. For instance, if you have set <code>image_data_format=channels_last</code>, then any model loaded from this repository will get built according to the TensorFlow data format convention, "Height-Width-Depth".</p>
<p>Note that:
- For <code>Keras < 2.2.0</code>, The Xception model is only available for TensorFlow, due to its reliance on <code>SeparableConvolution</code> layers.
- For <code>Keras < 2.1.5</code>, The MobileNet model is only available for TensorFlow, due to its reliance on <code>DepthwiseConvolution</code> layers.</p>
<hr />
<h2 id="usage-examples-for-image-classification-models">Usage examples for image classification models</h2>
<h3 id="classify-imagenet-classes-with-resnet50">Classify ImageNet classes with ResNet50</h3>
<pre><code class="python">from keras.applications.resnet50 import ResNet50
from keras.preprocessing import image
from keras.applications.resnet50 import preprocess_input, decode_predictions
import numpy as np
model = ResNet50(weights='imagenet')
img_path = 'elephant.jpg'
img = image.load_img(img_path, target_size=(224, 224))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
preds = model.predict(x)
# decode the results into a list of tuples (class, description, probability)
# (one such list for each sample in the batch)
print('Predicted:', decode_predictions(preds, top=3)[0])
# Predicted: [(u'n02504013', u'Indian_elephant', 0.82658225), (u'n01871265', u'tusker', 0.1122357), (u'n02504458', u'African_elephant', 0.061040461)]
</code></pre>
<h3 id="extract-features-with-vgg16">Extract features with VGG16</h3>
<pre><code class="python">from keras.applications.vgg16 import VGG16
from keras.preprocessing import image
from keras.applications.vgg16 import preprocess_input
import numpy as np
model = VGG16(weights='imagenet', include_top=False)
img_path = 'elephant.jpg'
img = image.load_img(img_path, target_size=(224, 224))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
features = model.predict(x)
</code></pre>
<h3 id="extract-features-from-an-arbitrary-intermediate-layer-with-vgg19">Extract features from an arbitrary intermediate layer with VGG19</h3>
<pre><code class="python">from keras.applications.vgg19 import VGG19
from keras.preprocessing import image
from keras.applications.vgg19 import preprocess_input
from keras.models import Model
import numpy as np
base_model = VGG19(weights='imagenet')
model = Model(inputs=base_model.input, outputs=base_model.get_layer('block4_pool').output)
img_path = 'elephant.jpg'
img = image.load_img(img_path, target_size=(224, 224))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
block4_pool_features = model.predict(x)
</code></pre>
<h3 id="fine-tune-inceptionv3-on-a-new-set-of-classes">Fine-tune InceptionV3 on a new set of classes</h3>
<pre><code class="python">from keras.applications.inception_v3 import InceptionV3
from keras.preprocessing import image
from keras.models import Model
from keras.layers import Dense, GlobalAveragePooling2D
from keras import backend as K
# create the base pre-trained model
base_model = InceptionV3(weights='imagenet', include_top=False)
# add a global spatial average pooling layer
x = base_model.output
x = GlobalAveragePooling2D()(x)
# let's add a fully-connected layer
x = Dense(1024, activation='relu')(x)
# and a logistic layer -- let's say we have 200 classes
predictions = Dense(200, activation='softmax')(x)
# this is the model we will train
model = Model(inputs=base_model.input, outputs=predictions)
# first: train only the top layers (which were randomly initialized)
# i.e. freeze all convolutional InceptionV3 layers
for layer in base_model.layers:
layer.trainable = False
# compile the model (should be done *after* setting layers to non-trainable)
model.compile(optimizer='rmsprop', loss='categorical_crossentropy')
# train the model on the new data for a few epochs
model.fit_generator(...)
# at this point, the top layers are well trained and we can start fine-tuning
# convolutional layers from inception V3. We will freeze the bottom N layers
# and train the remaining top layers.
# let's visualize layer names and layer indices to see how many layers
# we should freeze:
for i, layer in enumerate(base_model.layers):
print(i, layer.name)
# we chose to train the top 2 inception blocks, i.e. we will freeze
# the first 249 layers and unfreeze the rest:
for layer in model.layers[:249]:
layer.trainable = False
for layer in model.layers[249:]:
layer.trainable = True
# we need to recompile the model for these modifications to take effect
# we use SGD with a low learning rate
from keras.optimizers import SGD
model.compile(optimizer=SGD(lr=0.0001, momentum=0.9), loss='categorical_crossentropy')
# we train our model again (this time fine-tuning the top 2 inception blocks
# alongside the top Dense layers
model.fit_generator(...)
</code></pre>
<h3 id="build-inceptionv3-over-a-custom-input-tensor">Build InceptionV3 over a custom input tensor</h3>
<pre><code class="python">from keras.applications.inception_v3 import InceptionV3
from keras.layers import Input
# this could also be the output a different Keras model or layer
input_tensor = Input(shape=(224, 224, 3)) # this assumes K.image_data_format() == 'channels_last'
model = InceptionV3(input_tensor=input_tensor, weights='imagenet', include_top=True)
</code></pre>
<hr />
<h1 id="documentation-for-individual-models">Documentation for individual models</h1>
<table>
<thead>
<tr>
<th>Model</th>
<th align="right">Size</th>
<th align="right">Top-1 Accuracy</th>
<th align="right">Top-5 Accuracy</th>
<th align="right">Parameters</th>
<th align="right">Depth</th>
</tr>
</thead>
<tbody>
<tr>
<td><a href="#xception">Xception</a></td>
<td align="right">88 MB</td>
<td align="right">0.790</td>
<td align="right">0.945</td>
<td align="right">22,910,480</td>
<td align="right">126</td>
</tr>
<tr>
<td><a href="#vgg16">VGG16</a></td>
<td align="right">528 MB</td>
<td align="right">0.713</td>
<td align="right">0.901</td>
<td align="right">138,357,544</td>
<td align="right">23</td>
</tr>
<tr>
<td><a href="#vgg19">VGG19</a></td>
<td align="right">549 MB</td>
<td align="right">0.713</td>
<td align="right">0.900</td>
<td align="right">143,667,240</td>
<td align="right">26</td>
</tr>
<tr>
<td><a href="#resnet">ResNet50</a></td>
<td align="right">98 MB</td>
<td align="right">0.749</td>
<td align="right">0.921</td>
<td align="right">25,636,712</td>
<td align="right">-</td>
</tr>
<tr>
<td><a href="#resnet">ResNet101</a></td>
<td align="right">171 MB</td>
<td align="right">0.764</td>
<td align="right">0.928</td>
<td align="right">44,707,176</td>
<td align="right">-</td>
</tr>
<tr>
<td><a href="#resnet">ResNet152</a></td>
<td align="right">232 MB</td>
<td align="right">0.766</td>
<td align="right">0.931</td>
<td align="right">60,419,944</td>
<td align="right">-</td>
</tr>
<tr>
<td><a href="#resnet">ResNet50V2</a></td>
<td align="right">98 MB</td>
<td align="right">0.760</td>
<td align="right">0.930</td>
<td align="right">25,613,800</td>
<td align="right">-</td>
</tr>
<tr>
<td><a href="#resnet">ResNet101V2</a></td>
<td align="right">171 MB</td>
<td align="right">0.772</td>
<td align="right">0.938</td>
<td align="right">44,675,560</td>
<td align="right">-</td>
</tr>
<tr>
<td><a href="#resnet">ResNet152V2</a></td>
<td align="right">232 MB</td>
<td align="right">0.780</td>
<td align="right">0.942</td>
<td align="right">60,380,648</td>
<td align="right">-</td>
</tr>
<tr>
<td><a href="#inceptionv3">InceptionV3</a></td>
<td align="right">92 MB</td>
<td align="right">0.779</td>
<td align="right">0.937</td>
<td align="right">23,851,784</td>
<td align="right">159</td>
</tr>
<tr>
<td><a href="#inceptionresnetv2">InceptionResNetV2</a></td>
<td align="right">215 MB</td>
<td align="right">0.803</td>
<td align="right">0.953</td>
<td align="right">55,873,736</td>
<td align="right">572</td>
</tr>
<tr>
<td><a href="#mobilenet">MobileNet</a></td>
<td align="right">16 MB</td>
<td align="right">0.704</td>
<td align="right">0.895</td>
<td align="right">4,253,864</td>
<td align="right">88</td>
</tr>
<tr>
<td><a href="#mobilenetv2">MobileNetV2</a></td>
<td align="right">14 MB</td>
<td align="right">0.713</td>
<td align="right">0.901</td>
<td align="right">3,538,984</td>
<td align="right">88</td>
</tr>
<tr>
<td><a href="#densenet">DenseNet121</a></td>
<td align="right">33 MB</td>
<td align="right">0.750</td>
<td align="right">0.923</td>
<td align="right">8,062,504</td>
<td align="right">121</td>
</tr>
<tr>
<td><a href="#densenet">DenseNet169</a></td>
<td align="right">57 MB</td>
<td align="right">0.762</td>
<td align="right">0.932</td>
<td align="right">14,307,880</td>
<td align="right">169</td>
</tr>
<tr>
<td><a href="#densenet">DenseNet201</a></td>
<td align="right">80 MB</td>
<td align="right">0.773</td>
<td align="right">0.936</td>
<td align="right">20,242,984</td>
<td align="right">201</td>
</tr>
<tr>
<td><a href="#nasnet">NASNetMobile</a></td>
<td align="right">23 MB</td>
<td align="right">0.744</td>
<td align="right">0.919</td>
<td align="right">5,326,716</td>
<td align="right">-</td>
</tr>
<tr>
<td><a href="#nasnet">NASNetLarge</a></td>
<td align="right">343 MB</td>
<td align="right">0.825</td>
<td align="right">0.960</td>
<td align="right">88,949,818</td>
<td align="right">-</td>
</tr>
</tbody>
</table>
<p>The top-1 and top-5 accuracy refers to the model's performance on the ImageNet validation dataset.</p>
<p>Depth refers to the topological depth of the network. This includes activation layers, batch normalization layers etc.</p>
<hr />
<h2 id="xception">Xception</h2>
<pre><code class="python">keras.applications.xception.Xception(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
</code></pre>
<p>Xception V1 model, with weights pre-trained on ImageNet.</p>
<p>On ImageNet, this model gets to a top-1 validation accuracy of 0.790
and a top-5 validation accuracy of 0.945.</p>
<p>This model and can be built both with <code>'channels_first'</code> data format (channels, height, width) or <code>'channels_last'</code> data format (height, width, channels).</p>
<p>The default input size for this model is 299x299.</p>
<h3 id="arguments">Arguments</h3>
<ul>
<li>include_top: whether to include the fully-connected layer at the top of the network.</li>
<li>weights: one of <code>None</code> (random initialization) or <code>'imagenet'</code> (pre-training on ImageNet).</li>
<li>input_tensor: optional Keras tensor (i.e. output of <code>layers.Input()</code>) to use as image input for the model.</li>
<li>input_shape: optional shape tuple, only to be specified
if <code>include_top</code> is <code>False</code> (otherwise the input shape
has to be <code>(299, 299, 3)</code>.
It should have exactly 3 inputs channels,
and width and height should be no smaller than 71.
E.g. <code>(150, 150, 3)</code> would be one valid value.</li>
<li>pooling: Optional pooling mode for feature extraction
when <code>include_top</code> is <code>False</code>.<ul>
<li><code>None</code> means that the output of the model will be
the 4D tensor output of the
last convolutional block.</li>
<li><code>'avg'</code> means that global average pooling
will be applied to the output of the
last convolutional block, and thus
the output of the model will be a 2D tensor.</li>
<li><code>'max'</code> means that global max pooling will
be applied.</li>
</ul>
</li>
<li>classes: optional number of classes to classify images
into, only to be specified if <code>include_top</code> is <code>True</code>, and
if no <code>weights</code> argument is specified.</li>
</ul>
<h3 id="returns">Returns</h3>
<p>A Keras <code>Model</code> instance.</p>
<h3 id="references">References</h3>
<ul>
<li><a href="https://arxiv.org/abs/1610.02357">Xception: Deep Learning with Depthwise Separable Convolutions</a></li>
</ul>
<h3 id="license">License</h3>
<p>These weights are trained by ourselves and are released under the MIT license.</p>
<hr />
<h2 id="vgg16">VGG16</h2>
<pre><code class="python">keras.applications.vgg16.VGG16(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
</code></pre>
<p>VGG16 model, with weights pre-trained on ImageNet.</p>
<p>This model can be built both with <code>'channels_first'</code> data format (channels, height, width) or <code>'channels_last'</code> data format (height, width, channels).</p>
<p>The default input size for this model is 224x224.</p>
<h3 id="arguments_1">Arguments</h3>
<ul>
<li>include_top: whether to include the 3 fully-connected layers at the top of the network.</li>
<li>weights: one of <code>None</code> (random initialization) or <code>'imagenet'</code> (pre-training on ImageNet).</li>
<li>input_tensor: optional Keras tensor (i.e. output of <code>layers.Input()</code>) to use as image input for the model.</li>
<li>input_shape: optional shape tuple, only to be specified
if <code>include_top</code> is <code>False</code> (otherwise the input shape
has to be <code>(224, 224, 3)</code> (with <code>'channels_last'</code> data format)
or <code>(3, 224, 224)</code> (with <code>'channels_first'</code> data format).
It should have exactly 3 inputs channels,
and width and height should be no smaller than 32.
E.g. <code>(200, 200, 3)</code> would be one valid value.</li>
<li>pooling: Optional pooling mode for feature extraction
when <code>include_top</code> is <code>False</code>.<ul>
<li><code>None</code> means that the output of the model will be
the 4D tensor output of the
last convolutional block.</li>
<li><code>'avg'</code> means that global average pooling
will be applied to the output of the
last convolutional block, and thus
the output of the model will be a 2D tensor.</li>
<li><code>'max'</code> means that global max pooling will
be applied.</li>
</ul>
</li>
<li>classes: optional number of classes to classify images
into, only to be specified if <code>include_top</code> is <code>True</code>, and
if no <code>weights</code> argument is specified.</li>
</ul>
<h3 id="returns_1">Returns</h3>
<p>A Keras <code>Model</code> instance.</p>
<h3 id="references_1">References</h3>
<ul>
<li><a href="https://arxiv.org/abs/1409.1556">Very Deep Convolutional Networks for Large-Scale Image Recognition</a>: please cite this paper if you use the VGG models in your work.</li>
</ul>
<h3 id="license_1">License</h3>
<p>These weights are ported from the ones <a href="http://www.robots.ox.ac.uk/~vgg/research/very_deep/">released by VGG at Oxford</a> under the <a href="https://creativecommons.org/licenses/by/4.0/">Creative Commons Attribution License</a>.</p>
<hr />
<h2 id="vgg19">VGG19</h2>
<pre><code class="python">keras.applications.vgg19.VGG19(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
</code></pre>
<p>VGG19 model, with weights pre-trained on ImageNet.</p>
<p>This model can be built both with <code>'channels_first'</code> data format (channels, height, width) or <code>'channels_last'</code> data format (height, width, channels).</p>
<p>The default input size for this model is 224x224.</p>
<h3 id="arguments_2">Arguments</h3>
<ul>
<li>include_top: whether to include the 3 fully-connected layers at the top of the network.</li>
<li>weights: one of <code>None</code> (random initialization) or <code>'imagenet'</code> (pre-training on ImageNet).</li>
<li>input_tensor: optional Keras tensor (i.e. output of <code>layers.Input()</code>) to use as image input for the model.</li>
<li>input_shape: optional shape tuple, only to be specified
if <code>include_top</code> is <code>False</code> (otherwise the input shape
has to be <code>(224, 224, 3)</code> (with <code>'channels_last'</code> data format)
or <code>(3, 224, 224)</code> (with <code>'channels_first'</code> data format).
It should have exactly 3 inputs channels,
and width and height should be no smaller than 32.
E.g. <code>(200, 200, 3)</code> would be one valid value.</li>
<li>pooling: Optional pooling mode for feature extraction
when <code>include_top</code> is <code>False</code>.<ul>
<li><code>None</code> means that the output of the model will be
the 4D tensor output of the
last convolutional block.</li>
<li><code>'avg'</code> means that global average pooling
will be applied to the output of the
last convolutional block, and thus
the output of the model will be a 2D tensor.</li>
<li><code>'max'</code> means that global max pooling will
be applied.</li>
</ul>
</li>
<li>classes: optional number of classes to classify images
into, only to be specified if <code>include_top</code> is <code>True</code>, and
if no <code>weights</code> argument is specified.</li>
</ul>
<h3 id="returns_2">Returns</h3>
<p>A Keras <code>Model</code> instance.</p>
<h3 id="references_2">References</h3>
<ul>
<li><a href="https://arxiv.org/abs/1409.1556">Very Deep Convolutional Networks for Large-Scale Image Recognition</a></li>
</ul>
<h3 id="license_2">License</h3>
<p>These weights are ported from the ones <a href="http://www.robots.ox.ac.uk/~vgg/research/very_deep/">released by VGG at Oxford</a> under the <a href="https://creativecommons.org/licenses/by/4.0/">Creative Commons Attribution License</a>.</p>
<hr />
<h2 id="resnet">ResNet</h2>
<pre><code class="python">keras.applications.resnet.ResNet50(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
keras.applications.resnet.ResNet101(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
keras.applications.resnet.ResNet152(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
keras.applications.resnet_v2.ResNet50V2(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
keras.applications.resnet_v2.ResNet101V2(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
keras.applications.resnet_v2.ResNet152V2(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
</code></pre>
<p>ResNet, ResNetV2 models, with weights pre-trained on ImageNet.</p>
<p>This model and can be built both with <code>'channels_first'</code> data format (channels, height, width) or <code>'channels_last'</code> data format (height, width, channels).</p>
<p>The default input size for this model is 224x224.</p>
<h3 id="arguments_3">Arguments</h3>
<ul>
<li>include_top: whether to include the fully-connected layer at the top of the network.</li>
<li>weights: one of <code>None</code> (random initialization) or <code>'imagenet'</code> (pre-training on ImageNet).</li>
<li>input_tensor: optional Keras tensor (i.e. output of <code>layers.Input()</code>) to use as image input for the model.</li>
<li>input_shape: optional shape tuple, only to be specified
if <code>include_top</code> is <code>False</code> (otherwise the input shape
has to be <code>(224, 224, 3)</code> (with <code>'channels_last'</code> data format)
or <code>(3, 224, 224)</code> (with <code>'channels_first'</code> data format).
It should have exactly 3 inputs channels,
and width and height should be no smaller than 32.
E.g. <code>(200, 200, 3)</code> would be one valid value.</li>
<li>pooling: Optional pooling mode for feature extraction
when <code>include_top</code> is <code>False</code>.<ul>
<li><code>None</code> means that the output of the model will be
the 4D tensor output of the
last convolutional block.</li>
<li><code>'avg'</code> means that global average pooling
will be applied to the output of the
last convolutional block, and thus
the output of the model will be a 2D tensor.</li>
<li><code>'max'</code> means that global max pooling will
be applied.</li>
</ul>
</li>
<li>classes: optional number of classes to classify images
into, only to be specified if <code>include_top</code> is <code>True</code>, and
if no <code>weights</code> argument is specified.</li>
</ul>
<h3 id="returns_3">Returns</h3>
<p>A Keras <code>Model</code> instance.</p>
<h3 id="references_3">References</h3>
<ul>
<li><code>ResNet</code>: <a href="https://arxiv.org/abs/1512.03385">Deep Residual Learning for Image Recognition</a></li>
<li><code>ResNetV2</code>: <a href="https://arxiv.org/abs/1603.05027">Identity Mappings in Deep Residual Networks</a></li>
</ul>
<h3 id="license_3">License</h3>
<p>These weights are ported from the following:</p>
<ul>
<li><code>ResNet</code>: <a href="https://github.com/KaimingHe/deep-residual-networks">The original repository of Kaiming He</a> under the <a href="https://github.com/KaimingHe/deep-residual-networks/blob/master/LICENSE">MIT license</a>.</li>
<li><code>ResNetV2</code>: <a href="https://github.com/facebook/fb.resnet.torch">Facebook</a> under the <a href="https://github.com/facebook/fb.resnet.torch/blob/master/LICENSE">BSD license</a>.</li>
</ul>
<hr />
<h2 id="inceptionv3">InceptionV3</h2>
<pre><code class="python">keras.applications.inception_v3.InceptionV3(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
</code></pre>
<p>Inception V3 model, with weights pre-trained on ImageNet.</p>
<p>This model and can be built both with <code>'channels_first'</code> data format (channels, height, width) or <code>'channels_last'</code> data format (height, width, channels).</p>
<p>The default input size for this model is 299x299.</p>
<h3 id="arguments_4">Arguments</h3>
<ul>
<li>include_top: whether to include the fully-connected layer at the top of the network.</li>
<li>weights: one of <code>None</code> (random initialization) or <code>'imagenet'</code> (pre-training on ImageNet).</li>
<li>input_tensor: optional Keras tensor (i.e. output of <code>layers.Input()</code>) to use as image input for the model.</li>
<li>input_shape: optional shape tuple, only to be specified
if <code>include_top</code> is <code>False</code> (otherwise the input shape
has to be <code>(299, 299, 3)</code> (with <code>'channels_last'</code> data format)
or <code>(3, 299, 299)</code> (with <code>'channels_first'</code> data format).
It should have exactly 3 inputs channels,
and width and height should be no smaller than 75.
E.g. <code>(150, 150, 3)</code> would be one valid value.</li>
<li>pooling: Optional pooling mode for feature extraction
when <code>include_top</code> is <code>False</code>.<ul>
<li><code>None</code> means that the output of the model will be
the 4D tensor output of the
last convolutional block.</li>
<li><code>'avg'</code> means that global average pooling
will be applied to the output of the
last convolutional block, and thus
the output of the model will be a 2D tensor.</li>
<li><code>'max'</code> means that global max pooling will
be applied.</li>
</ul>
</li>
<li>classes: optional number of classes to classify images
into, only to be specified if <code>include_top</code> is <code>True</code>, and
if no <code>weights</code> argument is specified.</li>
</ul>
<h3 id="returns_4">Returns</h3>
<p>A Keras <code>Model</code> instance.</p>
<h3 id="references_4">References</h3>
<ul>
<li><a href="http://arxiv.org/abs/1512.00567">Rethinking the Inception Architecture for Computer Vision</a></li>
</ul>
<h3 id="license_4">License</h3>
<p>These weights are released under <a href="https://github.com/tensorflow/models/blob/master/LICENSE">the Apache License</a>.</p>
<hr />
<h2 id="inceptionresnetv2">InceptionResNetV2</h2>
<pre><code class="python">keras.applications.inception_resnet_v2.InceptionResNetV2(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
</code></pre>
<p>Inception-ResNet V2 model, with weights pre-trained on ImageNet.</p>
<p>This model and can be built both with <code>'channels_first'</code> data format (channels, height, width) or <code>'channels_last'</code> data format (height, width, channels).</p>
<p>The default input size for this model is 299x299.</p>
<h3 id="arguments_5">Arguments</h3>
<ul>
<li>include_top: whether to include the fully-connected layer at the top of the network.</li>
<li>weights: one of <code>None</code> (random initialization) or <code>'imagenet'</code> (pre-training on ImageNet).</li>
<li>input_tensor: optional Keras tensor (i.e. output of <code>layers.Input()</code>) to use as image input for the model.</li>
<li>input_shape: optional shape tuple, only to be specified
if <code>include_top</code> is <code>False</code> (otherwise the input shape
has to be <code>(299, 299, 3)</code> (with <code>'channels_last'</code> data format)
or <code>(3, 299, 299)</code> (with <code>'channels_first'</code> data format).
It should have exactly 3 inputs channels,
and width and height should be no smaller than 75.
E.g. <code>(150, 150, 3)</code> would be one valid value.</li>
<li>pooling: Optional pooling mode for feature extraction
when <code>include_top</code> is <code>False</code>.<ul>
<li><code>None</code> means that the output of the model will be
the 4D tensor output of the
last convolutional block.</li>
<li><code>'avg'</code> means that global average pooling
will be applied to the output of the
last convolutional block, and thus
the output of the model will be a 2D tensor.</li>
<li><code>'max'</code> means that global max pooling will
be applied.</li>
</ul>
</li>
<li>classes: optional number of classes to classify images
into, only to be specified if <code>include_top</code> is <code>True</code>, and
if no <code>weights</code> argument is specified.</li>
</ul>
<h3 id="returns_5">Returns</h3>
<p>A Keras <code>Model</code> instance.</p>
<h3 id="references_5">References</h3>
<ul>
<li><a href="https://arxiv.org/abs/1602.07261">Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning</a></li>
</ul>
<h3 id="license_5">License</h3>
<p>These weights are released under <a href="https://github.com/tensorflow/models/blob/master/LICENSE">the Apache License</a>.</p>
<hr />
<h2 id="mobilenet">MobileNet</h2>
<pre><code class="python">keras.applications.mobilenet.MobileNet(input_shape=None, alpha=1.0, depth_multiplier=1, dropout=1e-3, include_top=True, weights='imagenet', input_tensor=None, pooling=None, classes=1000)
</code></pre>
<p>MobileNet model, with weights pre-trained on ImageNet.</p>
<p>This model and can be built both with <code>'channels_first'</code> data format (channels, height, width) or <code>'channels_last'</code> data format (height, width, channels).</p>
<p>The default input size for this model is 224x224.</p>
<h3 id="arguments_6">Arguments</h3>
<ul>
<li>input_shape: optional shape tuple, only to be specified
if <code>include_top</code> is <code>False</code> (otherwise the input shape
has to be <code>(224, 224, 3)</code> (with <code>'channels_last'</code> data format)
or <code>(3, 224, 224)</code> (with <code>'channels_first'</code> data format).
It should have exactly 3 inputs channels,
and width and height should be no smaller than 32.
E.g. <code>(200, 200, 3)</code> would be one valid value.</li>
<li>alpha: controls the width of the network.<ul>
<li>If <code>alpha</code> < 1.0, proportionally decreases the number
of filters in each layer.</li>
<li>If <code>alpha</code> > 1.0, proportionally increases the number
of filters in each layer.</li>
<li>If <code>alpha</code> = 1, default number of filters from the paper
are used at each layer.</li>
</ul>
</li>
<li>depth_multiplier: depth multiplier for depthwise convolution
(also called the resolution multiplier)</li>
<li>dropout: dropout rate</li>
<li>include_top: whether to include the fully-connected
layer at the top of the network.</li>
<li>weights: <code>None</code> (random initialization) or
<code>'imagenet'</code> (ImageNet weights)</li>
<li>input_tensor: optional Keras tensor (i.e. output of
<code>layers.Input()</code>)
to use as image input for the model.</li>
<li>pooling: Optional pooling mode for feature extraction
when <code>include_top</code> is <code>False</code>.<ul>
<li><code>None</code> means that the output of the model
will be the 4D tensor output of the
last convolutional block.</li>
<li><code>'avg'</code> means that global average pooling
will be applied to the output of the
last convolutional block, and thus
the output of the model will be a
2D tensor.</li>
<li><code>'max'</code> means that global max pooling will
be applied.</li>
</ul>
</li>
<li>classes: optional number of classes to classify images
into, only to be specified if <code>include_top</code> is <code>True</code>, and
if no <code>weights</code> argument is specified.</li>
</ul>
<h3 id="returns_6">Returns</h3>
<p>A Keras <code>Model</code> instance.</p>
<h3 id="references_6">References</h3>
<ul>
<li><a href="https://arxiv.org/pdf/1704.04861.pdf">MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications</a></li>
</ul>
<h3 id="license_6">License</h3>
<p>These weights are released under <a href="https://github.com/tensorflow/models/blob/master/LICENSE">the Apache License</a>.</p>
<hr />
<h2 id="densenet">DenseNet</h2>
<pre><code class="python">keras.applications.densenet.DenseNet121(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
keras.applications.densenet.DenseNet169(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
keras.applications.densenet.DenseNet201(include_top=True, weights='imagenet', input_tensor=None, input_shape=None, pooling=None, classes=1000)
</code></pre>
<p>DenseNet models, with weights pre-trained on ImageNet.</p>
<p>This model and can be built both with <code>'channels_first'</code> data format (channels, height, width) or <code>'channels_last'</code> data format (height, width, channels).</p>
<p>The default input size for this model is 224x224.</p>
<h3 id="arguments_7">Arguments</h3>
<ul>
<li>blocks: numbers of building blocks for the four dense layers.</li>
<li>include_top: whether to include the fully-connected
layer at the top of the network.</li>
<li>weights: one of <code>None</code> (random initialization),
'imagenet' (pre-training on ImageNet),
or the path to the weights file to be loaded.</li>
<li>input_tensor: optional Keras tensor (i.e. output of <code>layers.Input()</code>)
to use as image input for the model.</li>
<li>input_shape: optional shape tuple, only to be specified
if <code>include_top</code> is False (otherwise the input shape
has to be <code>(224, 224, 3)</code> (with <code>'channels_last'</code> data format)
or <code>(3, 224, 224)</code> (with <code>'channels_first'</code> data format).
It should have exactly 3 inputs channels,
and width and height should be no smaller than 32.
E.g. <code>(200, 200, 3)</code> would be one valid value.</li>
<li>pooling: optional pooling mode for feature extraction
when <code>include_top</code> is <code>False</code>.<ul>
<li><code>None</code> means that the output of the model will be
the 4D tensor output of the
last convolutional block.</li>
<li><code>avg</code> means that global average pooling
will be applied to the output of the
last convolutional block, and thus
the output of the model will be a 2D tensor.</li>
<li><code>max</code> means that global max pooling will
be applied.</li>
</ul>
</li>
<li>classes: optional number of classes to classify images
into, only to be specified if <code>include_top</code> is True, and
if no <code>weights</code> argument is specified.</li>
</ul>
<h3 id="returns_7">Returns</h3>
<p>A Keras model instance.</p>
<h3 id="references_7">References</h3>
<ul>
<li><a href="https://arxiv.org/abs/1608.06993">Densely Connected Convolutional Networks</a> (CVPR 2017 Best Paper Award)</li>
</ul>
<h3 id="license_7">License</h3>
<p>These weights are released under <a href="https://github.com/liuzhuang13/DenseNet/blob/master/LICENSE">the BSD 3-clause License</a>.</p>
<hr />
<h2 id="nasnet">NASNet</h2>
<pre><code class="python">keras.applications.nasnet.NASNetLarge(input_shape=None, include_top=True, weights='imagenet', input_tensor=None, pooling=None, classes=1000)
keras.applications.nasnet.NASNetMobile(input_shape=None, include_top=True, weights='imagenet', input_tensor=None, pooling=None, classes=1000)
</code></pre>
<p>Neural Architecture Search Network (NASNet) models, with weights pre-trained on ImageNet.</p>
<p>The default input size for the NASNetLarge model is 331x331 and for the
NASNetMobile model is 224x224.</p>
<h3 id="arguments_8">Arguments</h3>
<ul>
<li>input_shape: optional shape tuple, only to be specified
if <code>include_top</code> is <code>False</code> (otherwise the input shape
has to be <code>(224, 224, 3)</code> (with <code>'channels_last'</code> data format)
or <code>(3, 224, 224)</code> (with <code>'channels_first'</code> data format)
for NASNetMobile or <code>(331, 331, 3)</code> (with <code>'channels_last'</code>
data format) or <code>(3, 331, 331)</code> (with <code>'channels_first'</code>
data format) for NASNetLarge.
It should have exactly 3 inputs channels,
and width and height should be no smaller than 32.
E.g. <code>(200, 200, 3)</code> would be one valid value.</li>
<li>include_top: whether to include the fully-connected
layer at the top of the network.</li>
<li>weights: <code>None</code> (random initialization) or
<code>'imagenet'</code> (ImageNet weights)</li>
<li>input_tensor: optional Keras tensor (i.e. output of
<code>layers.Input()</code>)
to use as image input for the model.</li>
<li>pooling: Optional pooling mode for feature extraction
when <code>include_top</code> is <code>False</code>.<ul>
<li><code>None</code> means that the output of the model
will be the 4D tensor output of the
last convolutional block.</li>
<li><code>'avg'</code> means that global average pooling
will be applied to the output of the
last convolutional block, and thus
the output of the model will be a
2D tensor.</li>
<li><code>'max'</code> means that global max pooling will
be applied.</li>
</ul>
</li>
<li>classes: optional number of classes to classify images
into, only to be specified if <code>include_top</code> is <code>True</code>, and
if no <code>weights</code> argument is specified.</li>
</ul>
<h3 id="returns_8">Returns</h3>
<p>A Keras <code>Model</code> instance.</p>
<h3 id="references_8">References</h3>
<ul>
<li><a href="https://arxiv.org/abs/1707.07012">Learning Transferable Architectures for Scalable Image Recognition</a></li>
</ul>
<h3 id="license_8">License</h3>
<p>These weights are released under <a href="https://github.com/tensorflow/models/blob/master/LICENSE">the Apache License</a>.</p>
<hr />
<h2 id="mobilenetv2">MobileNetV2</h2>
<pre><code class="python">keras.applications.mobilenet_v2.MobileNetV2(input_shape=None, alpha=1.0, include_top=True, weights='imagenet', input_tensor=None, pooling=None, classes=1000)
</code></pre>
<p>MobileNetV2 model, with weights pre-trained on ImageNet.</p>
<p>This model and can be built both with <code>'channels_first'</code> data format (channels, height, width) or <code>'channels_last'</code> data format (height, width, channels).</p>
<p>The default input size for this model is 224x224.</p>
<h3 id="arguments_9">Arguments</h3>
<ul>
<li>input_shape: optional shape tuple, to be specified if you would
like to use a model with an input img resolution that is not
(224, 224, 3).
It should have exactly 3 inputs channels (224, 224, 3).
You can also omit this option if you would like
to infer input_shape from an input_tensor.
If you choose to include both input_tensor and input_shape then
input_shape will be used if they match, if the shapes
do not match then we will throw an error.
E.g. <code>(160, 160, 3)</code> would be one valid value.</li>
<li>alpha: controls the width of the network. This is known as the
width multiplier in the MobileNetV2 paper.<ul>
<li>If <code>alpha</code> < 1.0, proportionally decreases the number
of filters in each layer.</li>
<li>If <code>alpha</code> > 1.0, proportionally increases the number
of filters in each layer.</li>
<li>If <code>alpha</code> = 1, default number of filters from the paper
are used at each layer.</li>
</ul>
</li>
<li>include_top: whether to include the fully-connected
layer at the top of the network.</li>
<li>weights: one of <code>None</code> (random initialization),
'imagenet' (pre-training on ImageNet),
or the path to the weights file to be loaded.</li>
<li>input_tensor: optional Keras tensor (i.e. output of
<code>layers.Input()</code>)
to use as image input for the model.</li>
<li>pooling: Optional pooling mode for feature extraction
when <code>include_top</code> is <code>False</code>.<ul>
<li><code>None</code> means that the output of the model
will be the 4D tensor output of the
last convolutional block.</li>
<li><code>'avg'</code> means that global average pooling
will be applied to the output of the
last convolutional block, and thus
the output of the model will be a
2D tensor.</li>
<li><code>'max'</code> means that global max pooling will
be applied. </li>
</ul>
</li>
<li>classes: optional number of classes to classify images
into, only to be specified if <code>include_top</code> is True, and
if no <code>weights</code> argument is specified.</li>
</ul>
<h3 id="returns_9">Returns</h3>
<p>A Keras model instance.</p>
<h3 id="raises">Raises</h3>
<p>ValueError: in case of invalid argument for <code>weights</code>,
or invalid input shape, alpha,
rows when weights='imagenet'</p>
<h3 id="references_9">References</h3>
<ul>
<li><a href="https://arxiv.org/abs/1801.04381">MobileNetV2: Inverted Residuals and Linear Bottlenecks</a></li>
</ul>
<h3 id="license_9">License</h3>
<p>These weights are released under <a href="https://github.com/tensorflow/models/blob/master/LICENSE">the Apache License</a>.</p>
</div>
</div>
<footer>
<div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
<a href="../backend/" class="btn btn-neutral float-right" title="Backend">Next <span class="icon icon-circle-arrow-right"></span></a>
<a href="../datasets/" class="btn btn-neutral" title="Datasets"><span class="icon icon-circle-arrow-left"></span> Previous</a>
</div>
<hr/>
<div role="contentinfo">
<!-- Copyright etc -->
</div>
Built with <a href="https://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" aria-label="versions">
<span class="rst-current-version" data-toggle="rst-current-version">
<a href="http://github.com/keras-team/keras/" class="fa fa-github" style="float: left; color: #fcfcfc"> GitHub</a>
<span><a href="../datasets/" style="color: #fcfcfc;">« Previous</a></span>
<span style="margin-left: 15px"><a href="../backend/" style="color: #fcfcfc">Next »</a></span>
</span>
</div>
<script>var base_url = '..';</script>
<script src="../js/theme.js" defer></script>
<script src="../search/main.js" defer></script>
<script type="text/javascript" defer>
window.onload = function () {
SphinxRtdTheme.Navigation.enable(true);
};
</script>
</body>
</html>
|