Action recognition in videos acquired by a moving camera using motion decomposition of Lagrangian particle trajectories

By Shandong Wu, Omar Oreifej, Mubarak Shah. 2011

183 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Histogram of Oriented Principal Components for Cross-View Action Recognition (2014, methodology): “Experimental results show that our method is able to achieve significantly better accuracy than current state-of-the-art methods for cross-view action recognition and is robust to action speed variations.”
HOPC: Histogram of Oriented Principal Components of 3D Pointclouds for Action Recognition (2014, neutral): “Motion trajectory based action recognition methods[30,34] are also not reliable in depth sequences [19].”
Efficient Pose-Based Action Recognition (2014, background): “For instance, [3, 4] rely majorly on motion cues to identify similar action sequences under static or moving camera setups.”
Better Exploiting Motion for Better Action Recognition (2013, background): “A rather different view is adopted in [32] where the motion decomposition is performed at the trajectory level.”
High-level event recognition in unconstrained videos (2012, methodology): “[159] used Lagrangian particle trajectories and decomposed the trajectories into camera-induced and object-induced components, which makes their method robust to camera motion.”
Learning a non-linear knowledge transfer model for cross-view action recognition (2015, background): “Especially, dense trajectories-based methods [30–32] have shown impressive results for action recognition by tracking densely sampled points using optical flow fields.”
HON4D: Histogram of Oriented 4D Normals for Activity Recognition from Depth Sequences (2013, methodology): “Additionally, standard methods for automatically acquiring motion trajectories in color images as in [25, 22] are also not reliable in depth sequences.”
Classifying web videos using a global video descriptor (2013, methodology): “Wu [12] captured ensemble motions of a scene on a set of dense Lagrangian particle trajectories, which were computed by numerical integration of optical flow over time.”
An Approach to Pose-Based Action Recognition (2013, methodology): “We recover F and C from M by rank minimization using the method described in [23].”
Dense Trajectories and Motion Boundary Descriptors for Action Recognition (2013, background): “6% and signiﬁcantly outperforms the state of the art (Wu et al, 2011b) by 5% .”
A Robust and Efficient Video Representation for Action Recognition (2015, unknown): “Wu et al (2011) apply a low-rank assumption to decompose feature trajectories into camera-induced and object-induced components.”
A Cause and Effect Analysis of Motion Trajectories for Modeling Actions (2014, methodology): “[29] used chaotic invariants on dense trajectories to classify actions after segmenting motion through Robust PCA (RPCA) subspace learning technique [16].”
Video-Based Human Behavior Understanding: A Survey (2013, methodology): “Dense optical flow has also been successfully used for action recognition [80], [81], [95].”
Human Action Recognition in Unconstrained Videos by Explicit Motion Modeling (2015, methodology): “In addition, Wang et al. [34] performed trajectory-based modeling using Bayesian models and Wu et al. [35] proposed to use decomposed Lagrangian particle trajectories for action recognition.”
Fuzzy human motion analysis: A review (2014, unknown): “] RA = 98.30 [208] ViHASi 2008 [209] - - RA = 72.00 [210] HOLLYWOOD 2008 [211] - - RA = 61.50 [212] HOLLYWOOD-2 2009 [213] - - RA = 64.30 [214] UCF-Sports 2008 [215] [164] RA = 85.77 [164] RA = 89.70 [216] UCF-11Youtube 2009 [217] - - RA = 89.79 [195] i3DPost 2009 [218] [163, 165, 164] RA = 100.00 [”
Learning a Deep Model for Human Action Recognition from Novel Viewpoints (2016, unknown): “[3], [4], spatio-temporal interest points [5], [6], [7], [8], [9], [10], and motion trajectories based representation [11], [12], [13], [14].”
A survey of video datasets for human action and activity recognition (2013, background): “The context makes reference, for example, to the information related to the set of actions compiled, the type of sceneries, the number of actors or the number of stored videos.”
Discovering discriminative action parts from mid-level video representations (2012, methodology): “The first uses bag-of-words representations, where the “words” are computed either statically for each frame [25, 14] or from trajectories [27, 19, 22, 26].”