ResNet

E74928

convolutional neural network architecture deep learning model residual network

ResNet is a deep convolutional neural network architecture known for its use of residual connections to enable very deep models and achieve state-of-the-art performance in image recognition tasks.

Aliases (9)

Deep Residual Learning for Image Recognition ×1
Pre-activation ResNet ×1
ResNet-101 ×1
ResNet-152 ×1
ResNet-18 ×1
ResNet-34 ×1
ResNet-50 ×1
ResNet-v2 ×1
Wide ResNet ×1

Statements (55)

Predicate	Object
instanceOf	convolutional neural network architecture → deep learning model → residual network →
achievedResult	state-of-the-art performance on ImageNet at introduction →
addresses	degradation problem in deep networks → vanishing gradient problem →
affiliationOfAuthors	Microsoft Research →
benchmarkedOn	CIFAR-10 → CIFAR-100 → ImageNet →
developedBy	Jian Sun → Kaiming He → Shaoqing Ren → Xiangyu Zhang →
domain	computer vision →
enables	training of very deep networks →
field	deep learning →
firstPublishedVenue	CVPR 2016 →
firstPublishedYear	2015 →
hasArchitectureDepth	101 layers → 152 layers → 18 layers → 34 layers → 50 layers →
hasConnectionType	identity skip connection → projection shortcut →
hasKeyIdea	identity mappings → residual learning → skip connections →
hasVariant	Pre-activation ResNet → ResNeXt → ResNet-101 → ResNet-152 → ResNet-18 → ResNet-34 → ResNet-50 → ResNet-v2 → Wide ResNet →
implementationAvailableIn	Keras → MXNet → PyTorch → TensorFlow →
influenced	design of modern computer vision backbones →
inspired	many subsequent CNN architectures →
introducedInPaper	Deep Residual Learning for Image Recognition →
optimizationBenefit	eases optimization of deep networks →
trainingMethod	stochastic gradient descent →
usedAs	backbone network in detection models →
usedFor	feature extraction → image classification → object detection →
usesComponent	ReLU activation → batch normalization → convolutional layer → residual block →

Referenced by (14)

Subject (surface form when different)	Predicate
ResNet ("ResNet-18") → ResNet ("ResNet-34") → ResNet ("ResNet-50") → ResNet ("ResNet-101") → ResNet ("ResNet-152") → ResNet ("Wide ResNet") → ResNet ("Pre-activation ResNet") → ResNet ("ResNet-v2") →	hasVariant
AlexNet → LeNet → VGG →	influenced
CLIP →	imageEncoderType
ResNet ("Deep Residual Learning for Image Recognition") →	introducedInPaper
torchvision →	modelFamily