Activity recognition using the velocity histories of tracked keypoints

By Ross Messing, C. Pal, Henry A. Kautz. IEEE International Conference on Computer Vision, 2009

523 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/14528cfb9f2f049cefa1c6bfef17b1f18110eac1

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

A Robust and Efficient Video Representation for Action Recognition (2015, unknown): “Feature trajectories (Matikainen et al, 2009; Messing et al, 2009; Sun et al, 2009; Wang et al, 2013a) have been shown to be a good way for capturing the intrinsic dynamics of video data.”
Inverse Dynamics for Action Recognition (2013, background): “2226879 [4], point trajectory [5], [6], and joint angle [7].”
Better Exploiting Motion for Better Action Recognition (2013, background): “More recent techniques assume some underlying temporal motion model involving trajectories [2, 6, 7, 17, 18, 25, 29, 32].”
Efﬁcient feature extraction, encoding and classiﬁcation for action recognition (unknown): “Several alternative local descriptors have been proposed including histograms of ﬂow orientations (HOF) [16], histograms of 3D gradients (HOG3D) [12, 33], motion boundary histograms (MBH) [4, 35], shapes of point trajectories [20, 26, 35], local trinary patterns [13, 37] and others.”
Jointly Learning Heterogeneous Features for RGB-D Activity Recognition (2015, background): “Bearing on these merits, it is believed that the introduced depth information can greatly light up the research of human activity analysis [12], [24], [36].”
A Data-Driven Approach for Event Prediction (2010, background): “Several works have proposed models to study, characterize, and classify human actions ranging from constrained environments [3, 4] to actions in the wild such as TV shows, sporting events, and cluttered backgrounds [5, 6].”
Going deeper into action recognition: A survey (2016, background): “In 2009, separate studies by Messing et al. (2009) and Matikainen et al. (2009) question the choice of ﬁxed shaped cuboids for action recognition and introduce the notion of trajectories.”
Dense Trajectories and Motion Boundary Descriptors for Action Recognition (2013, unknown): “To quantify the improvement obtained with our dense trajectories, we compare to three baseline trajectories in our experimental results: KLT trajectories (Matikainen et al, 2009; Messing et al, 2009; Sun et al, 2010), SIFT trajectories (Sun et al, 2009, 2010), as well as dense cuboids.”
“Important stuff, everywhere!” Activity recognition with salient proto-objects as context (2014, methodology): “However, the majority of works neglects contextual information and focuses on the recognition based on motion patterns alone [20, 18, 38, 42, 19].”
Modeling the Temporal Extent of Actions (2010, methodology): “’s University of Rochester Activities of Daily Living [20] and Marsaza lek et al .”
High-level event recognition in unconstrained videos (2012, background): “[87], who observed that velocity information is useful for detecting daily living actions in high-resolution videos.”
Understanding egocentric activities (2011, background): “Action recognition methods can be further categorized into multiple groups based on the features they use to represent actions: features based on the entire human figure [4, 5], local space-time features [14, 24], features based on interaction of objects and hands [9, 23] and features based on point”
Human Action Recognition in Unconstrained Videos by Explicit Motion Modeling (2015, methodology): “Instead of directly using the spatial-temporal local features in the bag-of-features representation, the other category performs temporal tracking of local patches and then computes features on top of the local patch trajectories [21], [22], [23], [24], [25], [26], [3].”
Script Data for Attribute-Based Recognition of Composite Activities (2012, background): “In consequence, most current research (with few exceptions [8, 3, 9]) focuses on simple basic-level activities such as walking or drinking, while the recognition of longer-term, complex, and composite activities such as assembling furniture or food preparation has been rarely addressed in computer v”
Efficient Feature Extraction, Encoding, and Classification for Action Recognition (2014, methodology): “While the recognition accuracy has been continuously improved, current methods remain limited to relatively small datasets due to the low speed of video processing, often ranging in the order of 1-2 frames per second.”
A survey of video datasets for human action and activity recognition (2013, unknown): “URADL 2009 [153] [131] Action recognition [131,2”
The Action Similarity Labeling Challenge (2012, background): “This was usually done for specific purposes, such as studying viewing variations [10], occlusions [25], or recognizing daily actions in static scenes [14].”
Actionness Ranking with Lattice Conditional Ordinal Random Fields (2014, background): “Notable examples include the introduction of local action features with bags-of-words framework [35], such as spatiotemporal interest points [21], trajectory-based representations [23, 34], and motion interchange patterns [17] and the more holistic action bank representation which embeds a video int”
A review on vision techniques applied to Human Behaviour Analysis for Ambient-Assisted Living (2012, methodology): “Velocity history of tracked key points is used in [65].”
A survey on vision-based human action recognition (2010, unknown): “[87]”