VGG-19

Info#

Only one version of VGG-19 has been built

@article{DBLP:journals/corr/SimonyanZ14a,
      author    = {Karen Simonyan and
                   Andrew Zisserman},
      title     = {Very Deep Convolutional Networks for Large-Scale Image Recognition},
      journal   = {CoRR},
      volume    = {abs/1409.1556},
      year      = {2014},
      url       = {http://arxiv.org/abs/1409.1556},
      timestamp = {Wed, 01 Oct 2014 15:00:05 +0200},
      biburl    = {http://dblp.uni-trier.de/rec/bib/journals/corr/SimonyanZ14a},
      bibsource = {dblp computer science bibliography, http://dblp.org}
    }

Keras Model Visulisation#

VGG-19

Keras Model Builds#

# Layer Cluster - 1
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(img_input)
x = Convolution2D(64, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(x)
x = Convolution2D(64, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = MaxPooling2D((2, 2), strides=(2, 2))(x)

# Layer Cluster - 2
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(x)
x = Convolution2D(128, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(x)
x = Convolution2D(128, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = MaxPooling2D((2, 2), strides=(2, 2))(x)

# Layer Cluster - 3
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(x)
x = Convolution2D(256, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(x)
x = Convolution2D(256, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(x)
x = Convolution2D(256, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(x)
x = Convolution2D(256, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = MaxPooling2D((2, 2), strides=(2, 2))(x)

# Layer Cluster - 4
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(x)
x = Convolution2D(512, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(x)
x = Convolution2D(512, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(x)
x = Convolution2D(512, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = ZeroPadding2D(padding=(1, 1), dim_ordering=DIM_ORDERING)(x)
x = Convolution2D(512, 3, 3,
                  activation='relu',
                  border_mode='same',
                  dim_ordering=DIM_ORDERING)(x)
x = MaxPooling2D((2, 2), strides=(2, 2))(x)

# Layer Cluster - 5 - Output Layer
x = Flatten()(x)
x = Dense(4096, activation='relu')(x)
x = Dropout(0.5)(x)
x = Dense(4096, activation='relu')(x)
x = Dropout(0.5)(x)
x = Dense(1000, activation='softmax')(x)