Motion Interchange Patterns for Action Recognition in Unconstrained Videos

By Orit Kliper-Gross, Yaron Gurovich, Tal Hassner, Lior Wolf. 2012

211 citations recorded by the source corpus.

Subfields: 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 Computer Vision and Pattern Recognition

Citations in the corpus

C3D: Generic Features for Video Analysis (2014, methodology): “Performance STIP [27] OSSML [28] MIP [26] MIP+STIP+MBH [16] iDT+FV [53] Random Chance Dataset .”
Bag of visual words and fusion methods for action recognition: Comprehensive study and good practice (2014, neutral): “[22] 2012 72.”
Learning Spatiotemporal Features with 3D Convolutional Networks (2014, methodology): “C3D Human Performance STIP [21] OSSML [22] MIP [20] MIP+STIP+MBH [11] iDT+FV [44] Imagenet Random Chance”
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 Critical Review of Action Recognition Benchmarks (2013, methodology): “68% in [10] with chance being 2% .”
Learning Spatio-Temporal Representation with Pseudo-3D Residual Networks (2017, unknown): “3% MIP [12] metric 65.”
From Actemes to Action: A Strongly-Supervised Representation for Detailed Action Understanding (2013, methodology): “We also compare with spacetime cuboids [5] on interest points, dense spacetime histogram of gradients (HoG3D) [14], and dense local motion interchange pattern [15].”
Motionlets: Mid-level 3D Parts for Human Motion Recognition (2013, methodology): “improvement over a recent method of action bank (highlevel representation), and 4 percents improvement over a recent feature of motion interchange pattern (low-level representation) [18].”
Large Margin Dimensionality Reduction for Action Similarity Labeling (2014, methodology): “Yair et al. [11] presented several variants of MIP, namely histMIP and DoGMIP, which replace the original patches employed in MIP by the Histogram of Gradient Orientations (HOG) and Difference of Gaussian (DoG), respectively.”
Mining Motion Atoms and Phrases for Complex Action Recognition (2013, methodology): “…Wang et al. [24] 77.2% Our best result 84.9% Methods Performance (accuracy) Laptev et al. [12] (from [20]) 47.9% Sadanand et al. [20] 57.9% Kliper-Gross et al. [11] 72.6% Reddy et al. [18] 76.9% Wang et al. [25] 78.4% Wang et al. [24] 85.6% Our best result 85.7% sentation for low-level features.”
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.”
Evaluating New Variants of Motion Interchange Patterns (2013, methodology): “In all our experiments we use the MIP parameters as described in [11].”
Learning to Share Latent Tasks for Action Recognition (2013, result): “We note that all these results obtained by using the same experiment setup in [22].”