Inception v4

E472889

Inception v4 is an advanced deep convolutional neural network model for image recognition that refines and extends earlier Inception architectures to achieve higher accuracy and efficiency.

Try in SPARQL Jump to: Statements Referenced by

Statements (46)

Predicate Object
instanceOf Inception architecture variant
convolutional neural network architecture
deep learning model
image classification model
achieves state-of-the-art accuracy on ImageNet at time of publication
architectureType deep convolutional neural network
basedOn Inception architecture NERFINISHED
benchmark ImageNet NERFINISHED
category feedforward neural network
contains multiple Inception-A blocks
multiple Inception-B blocks
multiple Inception-C blocks
reduction blocks between Inception stages
describedIn Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning NERFINISHED
designGoal computational efficiency
higher accuracy
developedBy Google NERFINISHED
Google Brain NERFINISHED
field computer vision
deep learning
follows Inception v3 NERFINISHED
implementationAvailableIn Keras NERFINISHED
TensorFlow NERFINISHED
improvesUpon GoogLeNet NERFINISHED
Inception v2 NERFINISHED
Inception v3 NERFINISHED
inputDomain natural images
inputType RGB images
license open source implementation available
networkDepth very deep
optimization batch normalization
paperAuthorsInclude Alex Alemi NERFINISHED
Christian Szegedy NERFINISHED
Sergey Ioffe NERFINISHED
Vincent Vanhoucke NERFINISHED
relatedTo Inception-ResNet-v1 NERFINISHED
Inception-ResNet-v2 NERFINISHED
task image classification
image recognition
trainingDataset ImageNet Large Scale Visual Recognition Challenge dataset NERFINISHED
typicalUseCase feature extraction from images
transfer learning for vision tasks
uses Inception modules
factorized convolutions
grid size reduction blocks
yearProposed 2016

Referenced by (1)

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

Inception architecture hasVariant Inception v4