VGG

E74406

VGG architecture variant convolutional neural network architecture deep learning model image classification model

VGG is a deep convolutional neural network architecture known for its simple, uniform use of small 3×3 filters and great depth, which achieved strong performance in image recognition tasks.

Jump to: Surface forms Statements Referenced by

Observed surface forms (5)

Surface form	Occurrences
VGG-16	1
VGG-19	1
VGG-11	1
VGG-13	1
VGGNet	1

Statements (50)

Predicate	Object
instanceOf	VGG architecture variant ⓘ VGG architecture variant ⓘ convolutional neural network architecture ⓘ deep learning model ⓘ image classification model ⓘ
achievedStateOfTheArtOn	ImageNet 2014 classification task ⓘ
category	computer vision model ⓘ
describedInPaper	Very Deep Convolutional Networks for Large-Scale Image Recognition ⓘ
designedFor	ImageNet ⓘ surface form: ImageNet Large Scale Visual Recognition Challenge image classification ⓘ image recognition ⓘ
developedAt	University of Oxford NERFINISHED ⓘ
developedBy	Visual Geometry Group ⓘ
field	computer vision ⓘ deep learning ⓘ
hasApplication	feature extraction for transfer learning ⓘ image retrieval ⓘ object recognition ⓘ
hasAuthor	Andrew Zisserman ⓘ Karen Simonyan ⓘ
hasCharacteristic	high parameter count ⓘ simple and uniform architecture ⓘ very deep network ⓘ
hasDesignPrinciple	increased network depth ⓘ uniform architecture across layers ⓘ use of small convolution filters ⓘ
hasInputImageSize	224×224 ⓘ
hasLimitation	computationally expensive ⓘ large memory usage ⓘ
hasNumberOfWeightLayers	16 ⓘ 19 ⓘ
hasVariant	VGG self-linksurface differs ⓘ surface form: VGG-11 VGG self-linksurface differs ⓘ surface form: VGG-13 VGG self-linksurface differs ⓘ surface form: VGG-16 VGG self-linksurface differs ⓘ surface form: VGG-19
implementedIn	Caffe ⓘ PyTorch ⓘ TensorFlow ⓘ
influenced	Inception architecture ⓘ surface form: Inception-based architectures ResNet ⓘ
introducedInYear	2014 ⓘ
isBaselineFor	many computer vision benchmarks ⓘ
paperArchive	arXiv:1409.1556 ⓘ
trainedOn	ImageNet ⓘ
usesActivationFunction	ReLU ⓘ
usesConvolutionFilterSize	1×1 ⓘ 3×3 ⓘ
usesFullyConnectedLayersAtEnd	true ⓘ
usesPoolingFilterSize	2×2 ⓘ
usesPoolingType	max pooling ⓘ

Referenced by (7)

Full triples — surface form annotated when it differs from this entity's canonical label.

VGG → hasVariant → VGG self-linksurface differs ⓘ

this entity surface form: VGG-11

VGG → hasVariant → VGG self-linksurface differs ⓘ

this entity surface form: VGG-13

VGG → hasVariant → VGG self-linksurface differs ⓘ

this entity surface form: VGG-16

VGG → hasVariant → VGG self-linksurface differs ⓘ

this entity surface form: VGG-19

AlexNet → influenced → VGG ⓘ

this entity surface form: VGGNet

LeNet → influenced → VGG ⓘ

torchvision (ecosystem) → modelFamily → VGG ⓘ

subject surface form: torchvision