Human Action Recognition in Videos Using Kinematic Features and Multiple Instance Learning

By Saad Ali, Mubarak Shah. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010

484 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/75dcf4f024c048668cd859b2974b69a433be9623

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

Joint segmentation and classification of human actions in video (2011, background): “Ali & Shah [1] recognized human actions using kinematic features.”
View-Invariant Action Recognition Based on Artificial Neural Networks (2012, background): “Actions are usually described by using either features based on motion information and optical flow [5], [6], or features devised mainly for action representation [7], [8].”
Online action recognition using covariance of shape and motion (2014, methodology): “Ali and Shah [7] derive kinematic features from optical flow and use them to represent spatio-temporal volumes in low dimensional space.”
The Moving Pose: An Efficient 3D Kinematics Descriptor for Low-Latency Action Recognition and Detection (2013, background): “Related Work: Traditional research on general action recognition focuses mainly on recognition accuracy using hidden Markov models, and more recently conditional random fields [14, 22], and less on reducing observational latency [1, 10, 20].”
Measuring and reducing observational latency when recognizing actions (2011, background): “Defining and measuring the observational latency of a system involves subtle decisions.”
Domain-Adaptive Discriminative One-Shot Learning of Gestures (2014, background): “TV broadcasts with aligned subtitles (or similarly actions from movies with aligned transcripts [1, 3, 14]).”
Learning to Separate Object Sounds by Watching Unlabeled Video (2018, background): “Max pooling in deep MIL [24,80,81] is typically used to identify the positive instances within an aggregated bag.”
Action recognition in videos acquired by a moving camera using motion decomposition of Lagrangian particle trajectories (2011, background): “Poisson equation-based features [6], 3D Haar feature [7]), spatiotemporal interest points ([8, 11, 14]), motion history image (MHI) [15], and kinematic features [12].”
Action recognition from motion capture data using Meta-Cognitive RBF Network classifier (2014, background): “Ali and Shah [18] recognized human actions using kinematic features.”
Evaluating New Variants of Motion Interchange Patterns (2013, background): “The optical flow between successive frames [1, 24], sub-volumes of the video [9], or surrounding the central motion [4, 5] is highly valuable for Action Recognition.”
Max-Margin Early Event Detectors (2012, methodology): “Most existing methods for event detection and modeling (e.g., Ke et al. 2005; Smith et al. 2005; Gorelick et al. 2007; Oh et al. 2008; Satkin and Hebert 2010; Klaser et al. 2010; Ali and Shah 2010; Niebles et al. 2010; Shi et al. 2010; Pei et al. 2011; Liu et al. 2011; Marin-Jiménez et al. 2011; Lan”
Challenges in multimodal gesture recognition (2016, background): “Along similar lines, (Ali and Shah, 2010) propose a set of kinematic features that are derived from the optical flow for human action recognition in videos: divergence, vorticity, symmetric and antisymmetric flow fields, second and third principal invariants of flow gradient and rate of strain…”
Inverse Dynamics for Action Recognition (2013, methodology): “optical flow-based approaches [3], [4], optical flow is used to derive a representation that is suitable for recognition.”
Object, Scene and Actions: Combining Multiple Features for Human Action Recognition (2010, unknown): “Ali and Shah use multiple-instance embedding [1] to facilitate classiﬁcation with their kinematic mode features.”
Motion Interchange Patterns for Action Recognition in Unconstrained Videos (2012, methodology): “g, [16]) or for whole video volumes as in [13] such information has been shown to provide strong cues to Action Recognition.”
High-level event recognition in unconstrained videos (2012, background): “, kinematic features [2,4]) is yet to be tested in the context of this problem.”
HyDR-MI: A hybrid algorithm to reduce dimensionality in multiple instance learning (2013, background): “There are approaches for text categorization (2), content-based image retrieval (30; 20) and image annotation (45; 31), drug activity prediction (27; 63), web index page recommendation (60; 52), semantic video retrieval (6), video concept detection (18; 1), pedestrian detection (29), and the predict”