Learning Optical Flow

By Deqing Sun, Stefan Roth, J. P. Lewis, Michael J. Black. European Conference on Computer Vision, 2008

302 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/ad61eebde119131f845dc902e1c8f7a4c6d66233

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

Optical flow modeling and computation: A survey (2015, methodology): “In [232], a learning approach is proposed for finding the best combination of constraints.”
Image phase shift invariance based cloud motion displacement vector calculation method for ultra-short-term solar PV power forecasting (2018, methodology): “The OF method calculates cloud pixel 74 displacement based on the assumption that the gray scale value of an image pixel remains constant during the cloud motion 75 [24,25], which is recently introduced in the research of solar forecasts for cloud velocity derivation [26,27].”
FlowNet: Learning Optical Flow with Convolutional Networks (2015, methodology): “[32] study statistics of optical flow and learn regularizers using Gaussian scale mixtures; Rosenbaum et al.”
Fully-trainable deep matching (2016, methodology): “[23] investigate the statistical properties of optical flow and learn the regularizers using Gaussian scale mixtures, Rosenbaum et al.”
Ieee Transactions on Pattern Analysis and Machine Intelligence 1 Large Displacement Optical Flow: Descriptor Matching in Variational Motion Estimation (2011, unknown): “For descriptors given on discrete grids in frame 1”
Layered image motion with explicit occlusions, temporal consistency, and depth ordering (2010, background): “Additionally, the parameters of the model could be learned [23], but this may require more extensive and representative training sets.”
PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume (2017, background): “[46] learn a full model for optical flow, but the learning has been limited to a few training sequences [4].”
Optical Flow Estimation Using a Spatial Pyramid Network (2016, background): “The network shows promising results, despite being trained on an artiﬁcial dataset of chairs ﬂying over randomly selected images.”
Secrets of optical flow estimation and their principles (2010, methodology): “530 HS [31] 35.”
A Database and Evaluation Methodology for Optical Flow (2007, methodology): “The training set includes the ground truth and is meant to be used for debugging, parameter estimation, and possibly even learning [41, 70].”
A Naturalistic Open Source Movie for Optical Flow Evaluation (2012, background): “The limited amount of non-test data included has meant that very few machine learning approaches have been applied to the flow problem (with a few exceptions [15]).”
Dense Semantic Correspondence Where Every Pixel is a Classifier (2015, methodology): “Since then, significant literature has focused on determining robust metrics under increasingly adverse conditions - from non-rigid deformations and occlusions, to non-global intensity, constrast and colorimetric changes [1, 14, 19, 20].”
StructureFlow: Image Inpainting via Structure-Aware Appearance Flow (2019, methodology): “Many methods [11, 27] have been proposed to accurately estimate optical flow between consecutive frames.”
Deep learning for physical processes: incorporating prior scientific knowledge (2017, methodology): “Classical methods rely on the brightness constancy constrain equation (BCCE) (equation 2), derived from the observation that surfaces usually persist over time and hence the intensity value of a small region remains the same despite its position change [Sun et al., 2008].”
Computing Range Flow from Multi-modal Kinect Data (2011, background): “1) should also work with most other global methods (like [8] or [13]).”
Motion Detail Preserving Optical Flow Estimation (2012, unknown): “In [33], Sun et al. empirically demonstrated that the simple Charbonnier function (' 1 -norm) actually outperforms other highly nonconvex robust functions due to its convex property.”
Super SloMo: High Quality Estimation of Multiple Intermediate Frames for Video Interpolation (2017, background): “Learning is often limited to a few parameters [13, 26, 30].”
A Quantitative Analysis of Current Practices in Optical Flow Estimation and the Principles Behind Them (2013, methodology): “For the coarse-to-ﬁne pyramid, Sun et al. (2008) use a downsampling factor of 0.8 during non-convex optimization.”