litlas
Log inCreate account

Convolutional Neural Network Architecture for Geometric Matching

Authors: Ignacio Rocco, Relja Arandjelovic, Josef Sivic - IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2017 cited by 514

We address the problem of determining correspondences between two images in agreement with a geometric model such as an affine, homography or thin-plate spline transformation, and estimating its parameters. The contributions of this work are three-fold. First, we propose a convolutional neural network architecture for geometric matching. The architecture is based on three main components that mimic the standard steps of feature extraction, matching and simultaneous inlier detection and model parameter estimation, while being trainable end-to-end. Second, we demonstrate that the network parameters can be trained from synthetically generated imagery without the need for manual annotation and that our matching layer significantly increases generalization capabilities to never seen before images. Finally, we show that the same model can perform both instance-level and category-level matching giving state-of-the-art results on the challenging PF, TSS and Caltech-101 datasets.

Distinctive ImageFeatures from…Distinctive Image Features from Scale-Invariant KeypointsDeformable SpatialPyramid Matching for…Deformable Spatial Pyramid Matching for Fast Dense CorrespondencesFlowWeb: Joint image setalignment by weaving…FlowWeb: Joint image set alignment by weaving consistent, pixel-wise correspondencesDiscriminative Learningof Deep Convolutional…Discriminative Learning of Deep Convolutional Feature Point DescriptorsProposal FlowProposal FlowWarpNet: WeaklySupervised Matching for…WarpNet: Weakly Supervised Matching for Single-View ReconstructionJoint Recovery of DenseCorrespondence and…Joint Recovery of Dense Correspondence and Cosegmentation in Two ImagesLearning DenseCorrespondence via…Learning Dense Correspondence via 3D-Guided Cycle ConsistencyLearning to Match AerialImages with Deep…Learning to Match Aerial Images with Deep Attentive ArchitecturesNetVLAD: CNNArchitecture for Weakly…NetVLAD: CNN Architecture for Weakly Supervised Place RecognitionDeepMatching:Hierarchical Deformable…DeepMatching: Hierarchical Deformable Dense MatchingProposal Flow: SemanticCorrespondences from…Proposal Flow: Semantic Correspondences from Object ProposalsDeep Fundamental MatrixEstimationDeep Fundamental Matrix EstimationDGC-Net: Dense GeometricCorrespondence NetworkDGC-Net: Dense Geometric Correspondence NetworkSPair-71k: A Large-scaleBenchmark for Semantic…SPair-71k: A Large-scale Benchmark for Semantic CorrespondenceSemantic Matching byWeakly Supervised 2D…Semantic Matching by Weakly Supervised 2D Point Set RegistrationD2-Net: A Trainable CNNfor Joint Description…D2-Net: A Trainable CNN for Joint Description and Detection of Local FeaturesClothFlow: A Flow-BasedModel for Clothed Perso…ClothFlow: A Flow-Based Model for Clothed Person GenerationHyperpixel Flow:Semantic Correspondence…Hyperpixel Flow: Semantic Correspondence With Multi-Layer Neural FeaturesWarp Consistency forUnsupervised Learning o…Warp Consistency for Unsupervised Learning of Dense CorrespondencesSelf-Supervised RigidRegistration for Small…Self-Supervised Rigid Registration for Small ImagesLearning ContrastiveRepresentation for…Learning Contrastive Representation for Semantic CorrespondencePDC-Net+: EnhancedProbabilistic Dense…PDC-Net+: Enhanced Probabilistic Dense Correspondence NetworkSemantic Correspondence:Unified Benchmarking an…Semantic Correspondence: Unified Benchmarking and a Strong BaselineConvolutional NeuralNetwork Architecture fo…Convolutional Neural Network Architecture for Geometric MatchingEarlier 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.