图像卷积网络模型汇总

New weights files: NASNet, DenseNet

 fchollet released this on 16 Jan

Assets12

• densenet121_weights_tf_dim_ordering_tf_kernels.h532.5 MB
• densenet121_weights_tf_dim_ordering_tf_kernels_notop.h528.6 MB
• densenet169_weights_tf_dim_ordering_tf_kernels.h557.1 MB
• densenet169_weights_tf_dim_ordering_tf_kernels_notop.h550.7 MB
• densenet201_weights_tf_dim_ordering_tf_kernels.h580.2 MB
• densenet201_weights_tf_dim_ordering_tf_kernels_notop.h572.9 MB
• NASNet-large-no-top.h5328 MB
• NASNet-large.h5343 MB
• NASNet-mobile-no-top.h519.1 MB
• NASNet-mobile.h523.1 MB
• Source code(zip)
• Source code(tar.gz)

Centralizing the storage location of weights files referred to in the keras.applications module.

•  v0.7
•  ccd0eb2

Add InceptionResNetV2

 fchollet released this on 9 Sep 2017

Assets4

• inception_resnet_v2_weights_tf_dim_ordering_tf_kernels.h5215 MB
• inception_resnet_v2_weights_tf_dim_ordering_tf_kernels_notop.h5209 MB
• Source code(zip)
• Source code(tar.gz)

v0.7

Add inception_resnet_v2

•  v0.6
•  37b252b

MobileNet

 fchollet released this on 1 Jul 2017 · 3 commits to master since this release

Assets34

• mobilenet_1_0_128_tf.h516.4 MB
• mobilenet_1_0_128_tf_no_top.h516.4 MB
• mobilenet_1_0_160_tf.h516.4 MB
• mobilenet_1_0_160_tf_no_top.h516.4 MB
• mobilenet_1_0_192_tf.h516.4 MB
• mobilenet_1_0_192_tf_no_top.h516.4 MB
• mobilenet_1_0_224_tf.h516.4 MB
• mobilenet_1_0_224_tf_no_top.h516.4 MB
• mobilenet_2_5_128_tf.h52.01 MB
• mobilenet_2_5_128_tf_no_top.h52.01 MB
• mobilenet_2_5_160_tf.h52.01 MB
• mobilenet_2_5_160_tf_no_top.h52.01 MB
• mobilenet_2_5_192_tf.h52.01 MB
• mobilenet_2_5_192_tf_no_top.h52.01 MB
• mobilenet_2_5_224_tf.h52.01 MB
• mobilenet_2_5_224_tf_no_top.h52.01 MB
• mobilenet_5_0_128_tf.h55.32 MB
• mobilenet_5_0_128_tf_no_top.h55.32 MB
• mobilenet_5_0_160_tf.h55.32 MB
• mobilenet_5_0_160_tf_no_top.h55.32 MB
• mobilenet_5_0_192_tf.h55.32 MB
• mobilenet_5_0_192_tf_no_top.h55.32 MB
• mobilenet_5_0_224_tf.h55.32 MB
• mobilenet_5_0_224_tf_no_top.h55.32 MB
• mobilenet_7_5_128_tf.h510.1 MB
• mobilenet_7_5_128_tf_no_top.h510.1 MB
• mobilenet_7_5_160_tf.h510.1 MB
• mobilenet_7_5_160_tf_no_top.h510.1 MB
• mobilenet_7_5_192_tf.h510.1 MB
• mobilenet_7_5_192_tf_no_top.h510.1 MB
• mobilenet_7_5_224_tf.h510.1 MB
• mobilenet_7_5_224_tf_no_top.h510.1 MB
• Source code(zip)
• Source code(tar.gz)

Code contributed by Somshubra Majumdar. Weights ported from tensorflow/models.

•  v0.5
•  63d19ac

Keras 2 API, new Inception V3 weights

 fchollet released this on 11 Mar 2017 · 6 commits to master since this release

Assets4

• inception_v3_weights_tf_dim_ordering_tf_kernels.h591.7 MB
• inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h583.8 MB
• Source code(zip)
• Source code(tar.gz)

v0.5

Update models to V2 API.

•  v0.4
•  48dcedb

Xception model

 fchollet released this on 20 Oct 2016 · 9 commits to master since this release

Assets4

• xception_weights_tf_dim_ordering_tf_kernels.h587.6 MB
• xception_weights_tf_dim_ordering_tf_kernels_notop.h579.8 MB
• Source code(zip)
• Source code(tar.gz)

Add the Xception model with ImageNet weights.

on 1 Oct 2016 

0.3 …

•  cc06786
•  
•  zip
•  
•  tar.gz

•  v0.3
•  cc06786

Music tagger model

 fchollet released this on 1 Oct 2016 · 12 commits to master since this release

Assets4

• music_tagger_crnn_weights_tf_kernels_tf_dim_ordering.h51.57 MB
• music_tagger_crnn_weights_tf_kernels_th_dim_ordering.h51.57 MB
• Source code(zip)
• Source code(tar.gz)

Add the music tagger CRNN by @keunwoochoi.

•  v0.2
•  3fdb8ce

InceptionV3, updated ResNet50

 fchollet released this on 10 Aug 2016 · 15 commits to master since this release

Assets10

• inception_v3_weights_tf_dim_ordering_tf_kernels.h590.7 MB
• inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h582.9 MB
• inception_v3_weights_th_dim_ordering_th_kernels.h590.7 MB
• inception_v3_weights_th_dim_ordering_th_kernels_notop.h582.9 MB
• resnet50_weights_tf_dim_ordering_tf_kernels.h598.1 MB
• resnet50_weights_tf_dim_ordering_tf_kernels_notop.h590.3 MB
• resnet50_weights_th_dim_ordering_th_kernels.h598.1 MB
• resnet50_weights_th_dim_ordering_th_kernels_notop.h590.3 MB
• Source code(zip)
• Source code(tar.gz)

Merge branch 'master' of https://github.com/fchollet/deep-learning-mo…

…dels

•  v0.1
•  1d56f9c

VGG16, VGG19, and ResNet50

 fchollet released this on 4 Aug 2016 · 19 commits to master since this release

Assets14

• resnet50_weights_tf_dim_ordering_tf_kernels.h598.1 MB
• resnet50_weights_tf_dim_ordering_tf_kernels_notop.h590.3 MB
• resnet50_weights_th_dim_ordering_th_kernels.h598.1 MB
• resnet50_weights_th_dim_ordering_th_kernels_notop.h590.3 MB
• vgg16_weights_tf_dim_ordering_tf_kernels.h5528 MB
• vgg16_weights_tf_dim_ordering_tf_kernels_notop.h556.2 MB
• vgg16_weights_th_dim_ordering_th_kernels.h5528 MB
• vgg16_weights_th_dim_ordering_th_kernels_notop.h556.2 MB
• vgg19_weights_tf_dim_ordering_tf_kernels.h5548 MB
• vgg19_weights_tf_dim_ordering_tf_kernels_notop.h576.4 MB
• vgg19_weights_th_dim_ordering_th_kernels.h5548 MB
• vgg19_weights_th_dim_ordering_th_kernels_notop.h576.4 MB
• Source code(zip)
• Source code(tar.gz)

Add code and weights for VGG16, VGG19, and ResNet50.

 

 

引：https://github.com/fchollet/deep-learning-models/releases