gptkbp:instanceOf
|
gptkb:convolutional_neural_network
|
gptkbp:activatedBy
|
gptkb:ReLU
|
gptkbp:architecture
|
sequential
|
gptkbp:author
|
gptkb:Andrew_Zisserman
gptkb:Karen_Simonyan
|
gptkbp:citation
|
over 100,000
|
gptkbp:competition
|
gptkb:ILSVRC_2014
|
gptkbp:developedBy
|
gptkb:University_of_Oxford
gptkb:Visual_Geometry_Group
|
gptkbp:finalLayerActivation
|
softmax
|
gptkbp:frameworksAvailableIn
|
gptkb:TensorFlow
gptkb:Keras
gptkb:PyTorch
|
gptkbp:hasBatchNormalization
|
no
|
gptkbp:hasPoolingLayers
|
yes
|
https://www.w3.org/2000/01/rdf-schema#label
|
VGG16
|
gptkbp:influenced
|
gptkb:Inception
gptkb:ResNet
gptkb:DenseNet
|
gptkbp:inputImageSize
|
224x224
|
gptkbp:introduced
|
gptkb:Very_Deep_Convolutional_Networks_for_Large-Scale_Image_Recognition
|
gptkbp:introducedIn
|
2014
|
gptkbp:level
|
16
|
gptkbp:license
|
gptkb:Creative_Commons_Attribution_License
|
gptkbp:notableFor
|
depth compared to previous models
simplicity of architecture
|
gptkbp:numberOfConvolutionalLayers
|
13
|
gptkbp:numberOfFullyConnectedLayers
|
3
|
gptkbp:openSource
|
yes
|
gptkbp:parameter
|
138 million
|
gptkbp:poolingType
|
max pooling
|
gptkbp:pretrainedWeightsAvailable
|
yes
|
gptkbp:successor
|
VGG19
|
gptkbp:top1ErrorRateOnImageNet
|
24.8%
|
gptkbp:top5ErrorRateOnImageNet
|
7.3%
|
gptkbp:trainer
|
gptkb:ImageNet
|
gptkbp:url
|
http://www.robots.ox.ac.uk/~vgg/research/very_deep/
|
gptkbp:usedFor
|
feature extraction
transfer learning
image classification
|
gptkbp:bfsParent
|
gptkb:Keras_Applications
gptkb:SegNet
|
gptkbp:bfsLayer
|
7
|