Efficient Deep Learning for Stereo Matching

By Wenjie Luo, Alexander Gerhard Schwing, Raquel Urtasun. 2016

777 citations recorded by the source corpus.

Subfields: Aerospace Engineering, 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 Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

RayNet: Learning Volumetric 3D Reconstruction with Ray Potentials (2018, methodology): “We follow common practice and remove the ReLU from the last layer in order to retain information encoded both in the negative and positive range [20].”
Fully-Convolutional Siamese Networks for Object Tracking (2016, methodology): “The same technique has been applied in contemporary work on stereo matching [23].”
Kornia: an Open Source Differentiable Computer Vision Library for PyTorch (2019, background): “Multi-view reconstruction is a well understood problem with a good geometric model [21], and many approaches for matching and optimization [39, 8, 51], and some recent promising deep learning approaches [34].”
Pyramid Stereo Matching Network (2018, background): “[18] propose a notably faster Siamese network in which the computation of matching costs is treated as a multi-label classification.”
Detail-Revealing Deep Video Super-Resolution (2017, methodology): “[22] used CNNs to learn a patch distance measure for stereo matching.”
HITNet: Hierarchical Iterative Tile Refinement Network for Real-time Stereo Matching (2020, background): “Early works trained siamese networks to extract patch-wise features or predict matching costs [35, 58, 56, 57].”
Group-Wise Correlation Stereo Network (2019, methodology): “The matching cost computation was accelerated in [13] by correlating unary features.”
DeepMVS: Learning Multi-view Stereopsis (2018, methodology): “For the use of ConvNets in visual reconstruction problems, early work focuses on learning patch similarity for stereo matching [42, 41, 26].”
DeepPruner: Learning Efficient Stereo Matching via Differentiable PatchMatch (2019, background): “They achieve state-of-the-art results by learning deep representations [22, 23] and building deep cost volumes [15, 8, 16].”
Computer Vision for Autonomous Vehicles: Problems, Datasets and State-of-the-Art (2017, methodology): “Recently, deep learning approaches (Žbontar & LeCun (2016); Luo et al. (2016); Mayer et al. (2016)) were proposed achieving state-of-the-art performance.”
Learning Depth with Convolutional Spatial Propagation Network (2018, methodology): “Following [28], several methods were proposed to increase computational efficiency [80], [81], or matching cost accuracy [82] with stronger network and confidence predictions.”
Context-Aware Single-Shot Detector (2017, methodology): “Most previous work utilized the theoretical receptive field sizes to guide their network designs [28, 21].”
A Survey on Deep Learning Techniques for Stereo-Based Depth Estimation (2020, methodology): “As such, correlation layers have been extensively used in the literature [22], [39], [41], [42], [44].”
Cascade Residual Learning: A Two-Stage Convolutional Neural Network for Stereo Matching (2017, positive): “In contrast to an independent binary classiﬁcation scheme between image patches, Luo et al. [19] proposed to learn a probability distribution over all disparity values.”
End-to-End Learning of Geometry and Context for Deep Stereo Regression (2017, background): “presented a notably faster network for computing local matching costs as a multi-label classification of disparities using a Siamese network [33].”
Unsupervised Monocular Depth Estimation with Left-Right Consistency (2016, background): “It has also been shown that posing this binocular correspondence search as a multi-class classification problem has advantages both in terms of quality of results and speed [38].”
Real-Time Self-Adaptive Deep Stereo (2018, methodology): “CNN based matching cost functions [44, 5, 20] achieved state-of-the-art on both KITTI and Middlebury v3 by replacing conventional cost functions [14] within the SGM pipeline [13].”
Deep Discrete Flow (2016, methodology): “to traditional local 3× 3 convolutional layers, we integrate context information by adopting dilated convolutions [18], which have recently demonstrated great performance in semantic segmentation.”
Fast Disparity Estimation Using Dense Networks (2018, methodology): “MC-CNN [4] concatenates the feature maps while Content-CNN [3] and Chen et al.”