FasterRCNN

E431008

FasterRCNN is a popular two-stage object detection architecture that first proposes candidate regions and then classifies and refines bounding boxes, widely used in computer vision tasks.

All labels observed (9)

How this entity was disambiguated

Statements (53)

Predicate Object
instanceOf computer vision model
convolutional neural network model
deep learning model
object detection architecture
two-stage detector
category region-based object detector
codeAvailableIn Caffe
PyTorch
TensorFlow
commonlyUsedIn autonomous driving perception
general object detection benchmarks
medical image analysis
surveillance systems
designedFor bounding box localization
object classification in images
object detection
evaluationMetric mean Average Precision
extends Fast R-CNN
linked to: R-CNN

R-CNN
fullName Faster Region-based Convolutional Neural Network
linked to: FasterRCNN
hasComponent ROI pooling layer
Region Proposal Network
linked to: FasterRCNN

backbone CNN
bounding box regression head
classification head
improvesOver Fast R-CNN
linked to: R-CNN

Selective Search region proposals
influenced Feature Pyramid Networks based detectors
Mask R-CNN
linked to: MaskRCNN
introducedInPaper Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
linked to: FasterRCNN
keyIdea learn region proposals with a neural network
share convolutional features between detection and proposal generation
output object class labels
refined bounding boxes
proposalType class-agnostic region proposals
proposedBy Jian Sun
Kaiming He
Ross Girshick
Shaoqing Ren
publishedAtConference NeurIPS 2015
linked to: NeurIPS
publishedYear 2015
stage1 Region Proposal Network
stage2 Fast R-CNN style detector
trainingDataset MS COCO
linked to: COCO

PASCAL VOC
typicalBackbone ResNet-101
linked to: ResNet

ResNet-50
linked to: ResNet

VGG16
linked to: VGG
uses Region Proposal Network
anchor boxes
multi-task loss
shared convolutional feature maps
stochastic gradient descent

How these facts were elicited

Referenced by (18)

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

torchvision modelFamily FasterRCNN
subject linked to: torchvision (ecosystem)
Kaiming He knownFor Faster R-CNN
linked to: FasterRCNN
Kaiming He notableWork Faster R-CNN (paper)
linked to: FasterRCNN
Kaiming He notableContribution advances in object detection with region-based CNNs
linked to: FasterRCNN
Shaoqing Ren knownFor Faster R-CNN
linked to: FasterRCNN
Shaoqing Ren coAuthorOf Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
linked to: FasterRCNN
Shaoqing Ren coDeveloperOf Faster R-CNN architecture
linked to: FasterRCNN
Shaoqing Ren contributedTo Region Proposal Networks
linked to: FasterRCNN
Shaoqing Ren notableWork Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
linked to: FasterRCNN
Jian Sun notablePublication Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
linked to: FasterRCNN
Faster R-CNN fullName Faster Region-based Convolutional Neural Network
subject linked to: FasterRCNN
linked to: FasterRCNN
Faster R-CNN introducedInPaper Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
subject linked to: FasterRCNN
linked to: FasterRCNN
Faster R-CNN hasComponent Region Proposal Network
subject linked to: FasterRCNN
linked to: FasterRCNN
Mask R-CNN basedOn Faster R-CNN
subject linked to: MaskRCNN
linked to: FasterRCNN
Mask R-CNN extends Faster R-CNN
subject linked to: MaskRCNN
linked to: FasterRCNN
Mask R-CNN improvesUpon Faster R-CNN
subject linked to: MaskRCNN
linked to: FasterRCNN
KeypointRCNN extends Faster R-CNN
linked to: FasterRCNN
DETR comparedTo Faster R-CNN
linked to: FasterRCNN