Layered image motion with explicit occlusions, temporal consistency, and depth ordering

By Deqing Sun, Erik B. Sudderth, Michael J. Black. 2010

151 citations recorded by the source corpus.

How researchers see this paper

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

Citations in the corpus

AutoFlow: Learning a Better Training Set for Optical Flow (2021, methodology): “We take a layered approach [49, 44] to rendering image pairs and their optical flow, as shown in Fig.”
Occlusion Aware Unsupervised Learning of Optical Flow (2017, background): “The last group builds more sophisticated frameworks such as modeling depth or a layered representation of objects to reason about occlusion [50, 48, 58, 43].”
Learning to Estimate Hidden Motions with Global Motion Aggregation (2021, background): “Since occlusions are caused by 3D motions, other works [35, 33] explicitly model local depth relationships into layers and reason about occlusions.”
Layered RGBD scene flow estimation (2015, methodology): “We assume that the scene consists of K independently moving layers, ordered in depth [35, 39].”
A Quantitative Analysis of Current Practices in Optical Flow Estimation and the Principles Behind Them (2013, background): “Keywords Optical flow estimation · Practices · Median filtering · Non-local term ·Motion boundary”
3D Scene Flow Estimation with a Piecewise Rigid Scene Model (2015, methodology): “More expressive priors have been introduced, for example, by virtue of an over-parameterization (Nir et al. 2008), layered (Sun et al. 2010) or piecewise planar scene models (Bleyer et al. 2011b).”
ClothCap (2017, methodology): “Here we focused on a model with three parts (body, pants and shirt), and future work should look at more garments, including estimating the number of garments automatically (as done in layered optical flow estimation (Sun et al. 2010)).”
Layered segmentation and optical flow estimation over time (2012, methodology): “Given the visibility mask, the occlusion reasoning (data likelihood term) is the same as in [27]”
SelFlow: Self-Supervised Learning of Optical Flow (2019, methodology): “A straightforward way is to add temporal constraints such as constant velocity [19, 22, 41], constant acceleration [45, 3], low-dimensional linear subspace [16], or rigid/non-rigid segmentation [48].”
EpicFlow: Edge-preserving interpolation of correspondences for optical flow (2015, background): “Their multi-label optimization can be interpreted as a motion segmentation problem or as a layered model [29].”
Competitive Collaboration: Joint Unsupervised Learning of Depth, Camera Motion, Optical Flow and Motion Segmentation (2018, background): “ntation network and state-of-the art ﬂow network [27], together with semantic segmentation should improve non-rigid segmentation. Additionally, the segmentation masks should be consistent across time [26]; this provides an additional strong constraint for unsupervised learning. For automotive applic”
Piecewise Rigid Scene Flow (2013, background): “[16] estimate general 2D motion by decomposition into several layers, which enables occlusion reasoning.”