Deformable Spatial Pyramid Matching for Fast Dense Correspondences

By Jaechul Kim, Ce Liu, Fei Sha, Kristen Grauman. 2013

266 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Learning to generate chairs with convolutional neural networks (2014, methodology): “In Table 1 we show the performance of our algorithm compared to human performance and two baselines: SIFT flow [21] and Deformable Spatial Pyramid [15] (DSP).”
Generative Visual Manipulation on the Natural Image Manifold (2016, methodology): “Additional evaluation: In the supplemental material, we evaluate our motion+color flow method, and compare our results against popular alignment methods that are designed to handle large displacement between two images [44,45].”
Self-Supervised Learning of Geometrically Stable Features Through Probabilistic Introspection (2018, background): “Yet, without any further training, they have been shown not to improve over hand-engineered features for geometric tasks [21, 8] and most approaches still combine hand-crafted features and spatial regularization [3, 13, 17, 19, 43].”
FlowWeb: Joint image set alignment by weaving consistent, pixel-wise correspondences (2015, methodology): “We compare our alignment performance with Congealing [23] (using SIFT), Collection Flow [18], DSP [21], and RASL [33].”
On the Link between Gaussian Homotopy Continuation and Convex Envelopes (2015, methodology): “Just within the past few years, it has been utilized for lowrank matrix recovery [45], error correction [48], super resolution [19], photometric stereo [72], image segmentation [35], face alignment [57], 3D surface estimation [1], motion estimation in videos [61], optical flow [10,62], shape and ill”
Learning to Generate Chairs, Tables and Cars with Convolutional Networks (2014, methodology): “n order to numerically evaluate the quality of the correspondences, we created a small test set of 30 image pairs (examples are shown in the supplementary material). We Method All Simple Difﬁcult DSP [15] 5:2 3:3 6:3 SIFT ﬂow [21] 4:0 2:8 4:8 Ours 3:9 3:9 3:9 Human 1:1 1:1 1:1 Table 1. Average displ”
Learning Correspondence From the Cycle-Consistency of Time (2019, methodology): “Given our focus on finding correspondence at the patch level, our method is also related to the classic SIFT Flow [39] algorithm and other methods for finding mid-level correspondences between regions across different scenes [30, 16, 87].”
Regularized Max Pooling for Image Categorization (2014, background): “[13] propose a pyramid graph model that simultaneously regularizes match consistency at multiple spatial extents.”
Convolutional Neural Network Architecture for Geometric Matching (2017, methodology): “We compare our method against SIFT Flow [37], Graph-matching kernels (GMK) [15], Deformable spatial pyramid matching (DSP) [31], DeepFlow [43], and all three variants of Proposal Flow (NAM, PHM, LOM) [23].”
WarpNet: Weakly Supervised Matching for Single-View Reconstruction (2016, methodology): “Reconstructions from VGG-M and DSP are noisy due to errors in matching and often produce extreme outlier points that had to be clipped for ease of visualization.”
Universal Correspondence Network (2016, methodology): “method SIFT-NN [22] HOG-NN [8] SIFT-flow [19] DaisyFF [32] DSP [18] DM best (1/2) [25] Ours-HN Ours-HN-ST MPI-Sintel 68.”
Learning Dense Correspondence via 3D-Guided Cycle Consistency (2016, methodology): “We also ran preliminary evaluation on a more recent follow-up based on deformable spatial pyramids [22], and found it to perform similarly to SIFT flow.”
Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network (2018, background): “In the beginning, there are a number of 2D facial alignment approaches which aim at locating a set of ﬁducial 2D facial landmarks, such as classic Active Appearance Model(AMM)[40,41,42] and Constrained Local Models(CLM)[43,44].”