Pose primitive based human action recognition in videos or still images

By Christian Thurau, Václav Hlaváč. 2008

339 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

A Hierarchical Pose-Based Approach to Complex Action Understanding Using Dictionaries of Actionlets and Motion Poselets (2016, background): “Initially, most research focused on pose-based action recognition in color videos [8, 27].”
On Recognizing Actions in Still Images via Multiple Features (2012, background): “Thurau and Hlavac [10] use non-negative matrix factorization on pose primitives, where the pose primitives are learnt from non-cluttered videos and applied to images for ﬁnding the closest pose.”
Learning Sparse Representations for Human Action Recognition (2012, neutral): “3 Thurau & Hlavac [32] 10 94.”
Social signal processing: Survey of an emerging domain (2009, neutral): “, [183][184]), optical flow (e.”
Recognizing actions by shape-motion prototype trees (2009, methodology): “In [25], an action is represented as a set of pose primitives and n-Gram models are used for action matching.”
HICO: A Benchmark for Recognizing Human-Object Interactions in Images (2015, background): “detecting human body parts) by deriving feature representations based on human poses [11, 38, 21, 30, 32, 31], whereas others focus more on spatial relations between humans and objects [34, 33, 23, 9, 5, 4].”
Understanding Video Events: A Survey of Methods for Automatic Interpretation of Semantic Occurrences in Video (2009, background): “Recently, the investigation of this domain has gained popularity partly because of the availability of public databases[40, 41], but largely because of the need to compare and contrast different abstraction schemes and their effect on event recognition performance [61, 62, 63, 64, 65].”
Discriminative Hierarchical Modeling of Spatio-temporally Composable Human Activities (2014, neutral): “However, traditional methods have several limitations: sillouhette based recognition assumes the camera is static [5, 31]; methods based on 2D body configurations require expensive body part detections to obtain initial input; etc.”
A Review of Human Activity Recognition Methods (2015, methodology): “A major focus in action recognition from still images or videos has been made in the context of scene appearance (Thurau and Hlavac, 2008; Yang et al., 2010;Maji et al., 2011).”
Efficient Pose-Based Action Recognition (2014, methodology): “Others utilized the human appearance as the basic building blocks in discriminating actions [5–7].”
Discovering Motion Primitives for Unsupervised Grouping and One-Shot Learning of Human Actions, Gestures, and Expressions (2013, neutral): “Other holistic representations of actions involve optical flow [32], spatiotemporal gradients [52], and HoG [39].”
Joint Angles Similarities and HOG2 for Action Recognition (2013, positive): “There have been several efforts for extending HOG to the IEEE Conference on Computer Vision and Pattern Recognition Workshops: Human Activity Understanding from 3D Data, to appear 2013 temporal domain [7] [13] [17] [9].”
Human Action Recognition With Video Data: Research and Evaluation Challenges (2014, background): “[86], [105], [134], [155] × × I I × × × × [24] S × I S × × × × [53], [75] × I × I × I √ × [52], [74], [78] × I × I × I × × [66] × S I I × × × × [69] × S I I × S √ × [129] × S I I × S × × [27] I S × I × S × × [14] I S × I × × × × [12] I I I × S × √ × [26], [30], [38], [42] I × I I × × √ × [35], [44] ”
Action recognition from a distributed representation of pose and appearance (2011, background): “Thurau and Hlavac [24] represent pose as a histogram of pose primitives.”
Weakly Supervised Learning of Interactions between Humans and Objects (2012, neutral): “The pose can, for example, be represented by a histogram-of-gradient (HOG) [7], [11] or based on shape correspondences [6].”
Coupled Action Recognition and Pose Estimation from Multiple Views (2012, background): “2005), previous works (Thurau and Hlavac 2008; Kläser et al. 2010; Natarajan et al. 2010; Yang et al. 2010) have encoded pose implicitly.”