• Publications
  • Influence
Fast Global Image Smoothing Based on Weighted Least Squares
TLDR
This paper presents an efficient technique for performing a spatially inhomogeneous edge-preserving image smoothing, called fast global smoother, focusing on sparse Laplacian matrices consisting of a data term and a prior term that approximate the solution of the memory- and computation-intensive large linear system by solving a sequence of 1D subsystems. Expand
Proposal Flow
TLDR
This work introduces a novel approach to semantic flow, dubbed proposal flow, that establishes reliable correspondences using object proposals, and demonstrates that proposal flow significantly outperforms existing semantic flow methods in various settings. Expand
Robust image filtering using joint static and dynamic guidance
TLDR
This work addresses the problem of how to transfer fine structures of guidance signals to input images, restoring noisy or altered structures in a data-dependent framework by jointly leveraging structural information of guidance and input images. Expand
Proposal Flow: Semantic Correspondences from Object Proposals
TLDR
This work introduces a novel approach to semantic flow, dubbed proposal flow, that establishes reliable correspondences using object proposals, and demonstrates that proposal flow significantly outperforms existing semantic flow methods in various settings. Expand
SCNet: Learning Semantic Correspondence
TLDR
A convolutional neural network architecture for learning a geometrically plausible model for semantic correspondence, called SCNet, that is trained on image pairs obtained from the PASCAL VOC 2007 keypoint dataset, and substantially outperforms both recent deep learning architectures and previous methods based on hand-crafted features. Expand
Robust Guided Image Filtering Using Nonconvex Potentials
TLDR
This work proposes a novel SD (for static/dynamic) filter that effectively controls the underlying image structure at different scales, and can handle a variety of types of data from different sensors, and has good edge-preserving smoothing properties. Expand
DASC: Dense adaptive self-correlation descriptor for multi-modal and multi-spectral correspondence
TLDR
A novel dense matching descriptor, called dense adaptive self-correlation (DASC), is proposed, based on the observation that a self-similarity existing within images is less sensitive to modality variations, and a randomized receptive field pooling, in which a sampling pattern is optimized with a discriminative learning. Expand
Learning Memory-Guided Normality for Anomaly Detection
TLDR
This work proposes to use a memory module with a new update scheme where items in the memory record prototypical patterns of normal data, boosting the discriminative power of both memory items and deeply learned features from normal data and lessening the representation capacity of CNNs. Expand
SFNet: Learning Object-Aware Semantic Correspondence
TLDR
A new CNN architecture is proposed, dubbed SFNet, which leverages a new and differentiable version of the argmax function for end-to-end training, with a loss that combines mask and flow consistency with smoothness terms. Expand
Space-Time Hole Filling With Random Walks in View Extrapolation for 3D Video
TLDR
A space-time joint filling method for color and depth videos in view extrapolation that is superior to state-of-the-art methods and provides both spatially and temporally consistent results with significantly reduced flicker artifacts. Expand
...
1
2
3
4
5
...