Learning human actions via information maximization

By Jingen Liu, Mubarak Shah. 2008 IEEE Conference on Computer Vision and Pattern Recognition, 2008

486 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/407c4513eaa073d32ea4e4ae21310a5327c97173

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

Cross-view action recognition via view knowledge transfer (2011, background): “Our conjecture is consistent with the idea of constructing semantic vocabularies for action recognition [20][26] by modelling human actions in a hierarchical manner.”
Recognising action as clouds of space-time interest points (2009, methodology): “Liu and Shah exploit the spatial distribution of interest points using a modified correlogram [10].”
Recognizing human actions by attributes (2011, methodology): “More recently, great progress has been made by introducing more descriptive action representations such as space-time pattern templates[1, 21], 2D shape matching [33, 7], optical flow patterns [6, 33], trajectory-based representation [27], and bag-of-video-words [24, 19, 20].”
Learning View-Invariant Sparse Representations for Cross-View Action Recognition (2013, background): “In order to accurately recognize human actions, most existing approaches focus on developing different discriminative features, such as spatio-temporal interest point (STIP) based features [32, 2, 13, 19], shape [16, 3, 21] and optical flow based features [5, 17, 16].”
Cross-View Action Recognition over Heterogeneous Feature Spaces (2013, methodology): “Some other methods [17, 3] learn a separate model for each action class in each view, however, it is difﬁcult to collect sufﬁcient labeled samples for each view to cover all the action classes.”
View-Invariant Action Recognition Using Latent Kernelized Structural SVM (2012, methodology): “Owing to the fact that there are only 12 actions completely conducted by all the 12 subjects, we compare our method using 12 actions and 12 subjects with the methods [32] [33] using 13 actions and 12 subjects.”
Cross-dataset action detection (2010, unknown): “Liu and Shah [12] 94.”
Exploring the Trade-off Between Accuracy and Observational Latency in Action Recognition (2012, methodology): “Bag-of-Words (BoW) is a straightforward approach that is known to consistently perform well on a wide variety of action datasets, such as Liu and Shah (2008). While Zhao and Elgammal (2008) propose extensions for the BoW model that use key frames, we use the original BoW model because recent work (M”
Deep Learning of Invariant Spatio-Temporal Features from Video (2010, methodology): “2% by Liu & Shah [21], used a bag-of-words approach on cuboid feature detectors and maximal mutual information to cluster the video-words.”
Human activity analysis (2011, background): “A large number of researchers have tested their system on the KTH dataset [Schuldt et al. 2004; Dollar et al. 2005; Jiang et al. 2006; Niebles et al. 2006; Yeo et al. 2006; Ke et al. 2007; Kim et al. 2007; Jhuang et al. 2007; Savarese et al. 2008; Laptev et al. 2008; Liu and Shah 2008; Bregonzio et ”
Video-Based Human Behavior Understanding: A Survey (2013, unknown): “points (STIPs) [86], [78], [98], [99], [100], [101], [102].”
Hidden Part Models for Human Action Recognition: Probabilistic versus Max Margin (2011, unknown): “8760 Liu & Shah [50] 0.”
Sequential Deep Learning for Human Action Recognition (2011, background): “[3] and others [12,17,2,4], which all use engineered motion and texture descriptors calculated around spatio-temporal interest points, which are manually engineered.”
Measuring and reducing observational latency when recognizing actions (2011, methodology): “We chose to use a Bag-of-Words (BoW) model for baseline comparison because this approach is straightforward and and has consistently performed well on a wide variety of action datasets, e.g., [10].”
Making Action Recognition Robust to Occlusions and Viewpoint Changes (2010, result): “9% recognition rate which is similar to results reported by other BoW based approaches [17, 14] on the IXMAS dataset.”