litlas
Log inCreate account

Neighbourhood Consensus Networks

Authors: Ignacio Rocco, Mircea Cimpoi, Relja Arandjelovic, Akihiko Torii, Tomás Pajdla, Josef Sivic - Conference on Neural Information Processing Systems, NeurIPS 2018 cited by 445

We address the problem of finding reliable dense correspondences between a pair of images. This is a challenging task due to strong appearance differences between the corresponding scene elements and ambiguities generated by repetitive patterns. The contributions of this work are threefold. First, inspired by the classic idea of disambiguating feature matches using semi-local constraints, we develop an end-to-end trainable convolutional neural network architecture that identifies sets of spatially consistent matches by analyzing neighbourhood consensus patterns in the 4D space of all possible correspondences between a pair of images without the need for a global geometric model. Second, we demonstrate that the model can be trained effectively from weak supervision in the form of matching and non-matching image pairs without the need for costly manual annotation of point to point correspondences. Third, we show the proposed neighbourhood consensus network can be applied to a range of matching tasks including both category- and instance-level matching, obtaining the state-of-the-art results on the PF Pascal dataset and the InLoc indoor visual localization benchmark.

An Iterative ImageRegistration Technique…An Iterative Image Registration Technique with an Application to Stereo VisionVideo Google: A TextRetrieval Approach to…Video Google: A Text Retrieval Approach to Object Matching in VideosDistinctive ImageFeatures from…Distinctive Image Features from Scale-Invariant KeypointsLocal Invariant FeatureDetectors: A SurveyLocal Invariant Feature Detectors: A SurveyVisualizing andUnderstanding…Visualizing and Understanding Convolutional NetworksDiscriminative Learningof Deep Convolutional…Discriminative Learning of Deep Convolutional Feature Point DescriptorsDeep Residual Learningfor Image RecognitionDeep Residual Learning for Image RecognitionPN-Net: Conjoined TripleDeep Network for…PN-Net: Conjoined Triple Deep Network for Learning Local Image DescriptorsComparative Evaluationof Hand-Crafted and…Comparative Evaluation of Hand-Crafted and Learned Local FeaturesGMS: Grid-Based MotionStatistics for Fast…GMS: Grid-Based Motion Statistics for Fast, Ultra-Robust Feature CorrespondenceSCNet: Learning SemanticCorrespondenceSCNet: Learning Semantic CorrespondenceInLoc: Indoor VisualLocalization With Dense…InLoc: Indoor Visual Localization With Dense Matching and View SynthesisDiscovering VisualPatterns in Art…Discovering Visual Patterns in Art Collections With Spatially-Consistent Feature LearningDiscoBox: WeaklySupervised Instance…DiscoBox: Weakly Supervised Instance Segmentation and Semantic Correspondence from Box SupervisionRetrieval andLocalization with…Retrieval and Localization with Observation ConstraintsCost Aggregation Is AllYou Need for Few-Shot…Cost Aggregation Is All You Need for Few-Shot SegmentationCost Aggregation with 4DConvolutional Swin…Cost Aggregation with 4D Convolutional Swin Transformer for Few-Shot SegmentationLepard: Learning partialpoint cloud matching in…Lepard: Learning partial point cloud matching in rigid and deformable scenesFS6D: Few-Shot 6D PoseEstimation of Novel…FS6D: Few-Shot 6D Pose Estimation of Novel ObjectsIntegrative Feature andCost Aggregation with…Integrative Feature and Cost Aggregation with Transformers for Dense CorrespondenceJoint Learning ofFeature Extraction and…Joint Learning of Feature Extraction and Cost Aggregation for Semantic CorrespondenceSemantic-AwareFine-Grained…Semantic-Aware Fine-Grained CorrespondenceNeural Matching Fields:Implicit Representation…Neural Matching Fields: Implicit Representation of Matching Fields for Visual CorrespondenceUnifying Feature andCost Aggregation with…Unifying Feature and Cost Aggregation with Transformers for Semantic and Visual CorrespondenceNeighbourhood ConsensusNetworksNeighbourhood Consensus NetworksEarlier referencesFocus paperCiting papersOlderNewer

Click a node to pin it, click the empty canvas to go back to this paper, or hover to preview. Open a node’s page from its title.