TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE

By S. Vedula, Simon Baker, P. Rander, Robert T. Collins, Takeo Kanade

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

Scene Flow to Action Map: A New Representation for RGB-D Based Action Recognition with Convolutional Neural Networks (2017, methodology): “With the advent of affordable RGB-D cameras, scene flow methods for RGB-D data have also been proposed [12, 31, 15, 16, 55, 39, 18, 51, 45].”
Occupancy Flow: 4D Reconstruction by Learning Particle Dynamics (2019, background): “Our representation is not only spatially and temporally continuous, but also implicitly provides correspondences at every point in space, so that OFlow can be seen as the continuous generalization of scene ﬂow [67,68].”
Live dense reconstruction with a single moving camera (2010, background): “In this case ∆xj is called scene flow [15]:”
Kinecting the dots: Particle based scene flow from depth sensors (2011, methodology): “[20] also bears some similarity to the techniques proposed in this paper.”
3D Scene Flow Estimation with a Piecewise Rigid Scene Model (2015, unknown): “Vedula et al. (1999) ﬁrst deﬁned scene ﬂow as the collective estimation of dense 3D geometry and 3D motion from image data.”
Stereoscopic Scene Flow Computation for 3D Motion Understanding (2011, background): “Three-dimensional scene flow (Vedula et al. 2005) 17 Yes Close ?”
Optical flow modeling and computation: A survey (2015, unknown): “The problem of recovering the 3D motion itself from several synchronized cameras is also an active research domain known as scene ﬂow estimation [18,248].”
Total Moving Face Reconstruction (2014, methodology): “Similar to scene flow methods [39], we reconstruct a dense 3D flow field; key differences include our illumination invariance model, and that we compute 3D to 2D correspondence rather than 3D to 3D.”
Spacetime stereo: shape recovery for dynamic scenes (2003, methodology): “[18] present a linear algorithm to compute 3D scene flow based on 2D optical flow and estimate 3D structures from the scene flow.”
A primal-dual framework for real-time dense RGB-D scene flow (2015, unknown): “[3] who proposed to compute the Lucas-Kanade optical flow [4] first and apply the range flow constraint equation at a later stage, obtaining a local solution which does not exploit the geometric data to estimate the optical flow.”
On 3D scene flow and structure estimation (2001, unknown): “Vedula et al. [ 19 ] designed linear algorithms to compute 3D scene flow under three different scenarios.”
SphereFlow: 6 DoF Scene Flow from RGB-D Pairs (2014, background): “[26]—by flowing all 3D points in both frames for our final results on the data set in Figure 9 to intermediate points in time, demonstrating the visual credibility of our recovered motions.”
Dense Semi-rigid Scene Flow Estimation from RGBD Images (2014, methodology): “Scene flow was first introduced by Vedula [18] as the 3D motion field of the scene.”
Spacetime faces: high resolution capture for modeling and animation (2004, methodology): “…the consistency of the optical (cid:3)ow and the vertex inter-frame motion U n ; t = d n ; t + 1 (cid:0) d n ; t , called scene (cid:3)ow in [Vedula et al. 1999], by the following metric: where p k ( U n ; t ) is the image projection of U n ; t in the k ’th image plane and u n ; t ; k is…”
Learning Motion Patterns in Videos (2016, methodology): “, 3D motion field in a scene [37], is another form of motion estimation, but is computed with additional information, such as disparity values computed from stereo images [15, 40], or estimated 3D scene models [38].”