Beyond pixels: exploring new representations and applications for motion analysis

By Ce Liu. 2009

928 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

The Unreasonable Effectiveness of Deep Features as a Perceptual Metric (2018, methodology): “Frame interpolation We sample patches from different frame interpolation algorithms, including three variants of flow-based interpolation [33], CNN-based interpolation [39], and phase-based interpolation [37] on the Davis Middleburry dataset [49].”
Unsupervised Learning of Disentangled Representations from Video (2017, background): “ansformations of groups of pixels. Cricri et al. [3] use a ladder of stacked-autoencoders. Other works predict optical ﬂows ﬁelds that can be used to extrapolate motion beyond the current frame, e.g. [16, 38, 33]. In contrast, a single pose vector is predicted in our model, rather than a spatial ﬁel”
Stochastic Video Generation with a Learned Prior (2018, background): “Other works predict optical ﬂows ﬁelds that can be used to extrapolate motion beyond the current frame, e.g. ( Liu, 2009; Xue et al., 2016; Walker et al., 2015).”
Tracking Emerges by Colorizing Videos (2018, methodology): “cluding network architecture. 8 Vondrick, Shrivastava, Fathi, Guadarrama, Murphy Method Supervised? Segment Boundary Identity 22.1 23.6 Single Image Colorization 4.7 5.2 Optical Flow (Coarse-to-Fine) [59] 13.0 15.1 Optical Flow (FlowNet2) [23] 26.7 25.2 Ours 34.6 32.7 Fully Supervised [46,47] X 55.1”
Deep Stereo: Learning to Predict New Views from the World's Imagery (2015, methodology): “Figure 8: Comparison of real to synthesized KITTI images, showing results from the Kitti- and Street-View-trained model, a depth splatting algorithm from computed depth maps [29] and an optical flow algorithm [26].”
Cross Pixel Optical Flow Similarity for Self-Supervised Learning (2018, neutral): “We extract 8 random frames from each video and compute optical-flow between those at times t and t 5 using the same (handcrafted) optical-flow method of [42, 33].”
Video magnification in presence of large motions (2015, methodology): “For Lagrangian, After Effects and Liu [6] we only mag-nify the Region of Interest.”
Motion-based counter-measures to photo attacks in face recognition (2014, methodology): “For our experiments in this article, we have chosen to use Liu’s [23] implementation for estimating dense optical flow5, as it is developed in C++, it executes in much faster pace and could be easily ported into our unified framework.”
Detecting anomalous events in videos by learning deep representations of appearance and motion (2017, methodology): “The code for optical flow calculation is written in C++ and wrapped with Matlab mex for computational efficiency [51].”
Video Object Segmentation using Teacher-Student Adaptation in a Human Robot Interaction (HRI) Setting (2018, methodology): “The input to the motion stream is the optical flow computed using [16], and converted into RGB representation using the Sintel color wheel representation [1].”
Video Segmentation by Non-Local Consensus voting (2014, methodology): “to a low resolution (size 100X100) using the efficient code of [22] (runtime is less than a second per frame).”
Crowded Scene Analysis: A Survey (2015, methodology): “Commencing with the optical ﬂow ﬁeld estimation adapted from [94], Loy et al. [95] presented a global motion saliency detection framework.”
Violent flows: Real-time detection of violent crowd behavior (2012, methodology): “Our method was implemented in MATLAB, the opticalflow code available from [22], and linear SVM [7].”
Learning Deep Representations of Appearance and Motion for Anomalous Event Detection (2015, methodology): “…MPPCA[10]Social Force + MPPCA[16] Social Force[17] Sparse reconstruction[4] Local statistical[22] Detection at 150FPS[15] AMDN MDT[16]MPPCA[10]Social Force + MPPCA[16] Social Force[17] Sparse Reconstruction[4] Detection at 150FPS[15] The Train dataset [36] depicts moving people inside a train.”
AMP: Adaptive Masked Proxies for Few-Shot Segmentation (2019, methodology): “[15] and converted to RGB using a color wheel.”
Depth Transfer: Depth Extraction from Video Using Non-Parametric Sampling (2014, methodology): “For further details and discussion of IRLS, see the appendix of Liu’s thesis [31].”