Fast Edge-Preserving PatchMatch for Large Displacement Optical Flow

By Linchao Bao, Qingxiong Yang, Hailin Jin. 2014

184 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

FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks (2016, methodology): “(3)An exception is EPPM for which we could not provide the required Windows environment and use the results from [4].”
FlowNet: Learning Optical Flow with Convolutional Networks (2015, methodology): “After fine-tuning on Sintel our networks can outperform the competing real-time method EPPM [3] on Sintel Final and KITTI while being twice as fast.”
Weakly Supervised Semantic Segmentation Using Web-Crawled Videos (2017, unknown): “We define a temporal edge between two superpixels from consecutive frames if they are connected by at least one optical flow [1].”
Unsupervised Deep Learning for Optical Flow Estimation (2017, methodology): “Method Chairs KITTI2012 KITTI2015 Sintel Clean Sintel Final Time (Sec) testing train test train test train test train test CPU GPUocc noc occ noc occ noc fl-all fl-all occ noc occ noc occ noc occ noc Epicflow 2.94 3.47 1.48 3.8 1.5 9.57 4.45 0.28 0.27 2.40 1.23 4.12 1.36 3.70 2.63 6.29 3.06 16 — Dee”
Massively Parallel Multiview Stereopsis by Surface Normal Diffusion (2015, background): “The resulting low memory requirements (independent of the disparity range) make Patchmatch Stereo well-suited for large images or memory-constrained environments, including implementation on GPU which modify the original sequential propagation scheme [1, 2, 17, 40].”
Occlusion Aware Unsupervised Learning of Optical Flow (2017, background): “Another research direction aims on improving the computational complexity for optical flow methods [53, 10, 24], since the energy minimization problem over high resolution images usually require a large computational complexity.”
Fast Optical Flow Using Dense Inverse Search (2016, methodology): “Recently, only very few works aimed at balancing accuracy and run-time in favor of efﬁciency [17, 48, 54], or employed massively parallelized dedicated hardware to achieve acceptable run-times [4, 40, 18].”
Flow Fields: Dense Correspondence Fields for Highly Accurate Large Displacement Optical Flow Estimation (2015, methodology): “Bao et al. [3] used an edge aware bilateral data term instead.”
DenseNet for dense flow (2017, methodology): “Method Chairs Sintel KITTI Runtime EPPM [18] − 8.”
EpicFlow: Edge-preserving interpolation of correspondences for optical flow (2015, result): “In contrast, our approach is conceptually closer to recent work that rely mainly on descriptor matching [36, 23, 12, 22, 37, 27, 5].”
SegFlow: Joint Learning for Video Object Segmentation and Optical Flow (2017, unknown): “EPPM [3] - 6.”