UNet architecture

GPTKB entity

Statements (52)
Predicate Object
gptkbp:instanceOf deep learning model architecture
gptkbp:activatedBy gptkb:ReLU
gptkbp:architecture gptkb:convolutional_neural_network
gptkbp:commonIn biomedical image analysis
medical image segmentation
gptkbp:firstPublished gptkb:U-Net:_Convolutional_Networks_for_Biomedical_Image_Segmentation
gptkbp:frameworksAvailableIn gptkb:TensorFlow
gptkb:Keras
gptkb:PyTorch
gptkbp:hasBatchNormalization optional
gptkbp:hasContractingPath yes
gptkbp:hasConvolutionalLayers yes
gptkbp:hasDecoder yes
gptkbp:hasDownsamplingLayers yes
gptkbp:hasDropout optional
gptkbp:hasEncoder yes
gptkbp:hasExpandingPath yes
gptkbp:hasPoolingLayers yes
gptkbp:hasSkipConnections yes
gptkbp:hasSymmetricStructure yes
gptkbp:hasTransposedConvolutions yes
gptkbp:hasUpsamplingLayers yes
gptkbp:hasVariant gptkb:3D_U-Net
gptkb:Attention_U-Net
gptkb:ResUNet
gptkb:UNet++
gptkb:MultiResUNet
gptkb:Nested_U-Net
gptkb:U2-Net
gptkb:UNet3+
gptkb:V-Net
https://www.w3.org/2000/01/rdf-schema#label UNet architecture
gptkbp:input gptkb:illustrator
gptkbp:inputShape variable
gptkbp:inspiredBy gptkb:fully_convolutional_networks
gptkbp:introduced gptkb:Olaf_Ronneberger
gptkbp:introducedIn 2015
gptkbp:lossFunction cross-entropy
dice loss
gptkbp:notableFor precise localization
works with few training images
end-to-end training
efficient use of training data
gptkbp:openSource yes
gptkbp:output segmentation map
gptkbp:outputActivationFunction sigmoid
softmax
gptkbp:outputShape variable
gptkbp:usedFor image segmentation
gptkbp:bfsParent gptkb:Stable_Diffusion_2.0
gptkb:Stable_Diffusion_model
gptkbp:bfsLayer 6