Action recognition by learning mid-level motion features

By Alireza Fathi, Greg Mori. 2008

538 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Biomedical Engineering, 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 Biomedical Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Delving into Egocentric Actions (2015, background): “Fathi and Mori [8] propose to construct mid-level motion features from optical flow using AdaBoost.”
Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset (2017, methodology): “In our model comparisons, we did not consider more classic approaches such as bag-of-visual-words representations [6, 19, 22, 33].”
A Spatio-Temporal Descriptor Based on 3D-Gradients (2008, unknown): “Note that we cannot compare to results given by Jhuang et al. [7] or Fathi and Mori [5].”
Trajectons: Action recognition through the motion analysis of tracked features (2009, methodology): “We propose two variants of trajectons, which differ in their construction of trajectory snippets.”
Motion Interchange Patterns for Action Recognition in Unconstrained Videos (2012, methodology): “These methods first estimate the optical-flow between successive frames [13, 14], sub-volumes of the whole video [15], or surrounding the central motion [16, 17].”
Going Deeper into First-Person Activity Recognition (2016, background): “proposed mid-level motion features and gaze for recognizing ego-centric activities in [6, 7].”
Recognizing actions by shape-motion prototype trees (2009, methodology): “Feature extraction methods can be roughly classified into four categories: motion-based [6] [8] [29] [30], appearancebased [7, 25], space-time volume-based [2, 10, 13], and space-time interest points and local feature-based [5, 12, 19, 21, 24].”
Progressive Teacher-Student Learning for Early Action Prediction (2019, background): “Some representative handcrafted features like Cuboids [7] [44] [6], interest point clouds [4], 3DHOG [20], SIFT [33], and dense trajectory [41] etc.”
Recognising action as clouds of space-time interest points (2009, methodology): “Flow based approaches construct action templates based on optical flow computation [4, 5].”
Local Trinary Patterns for human action recognition (2009, background): “First, there are contributions which compute representations for nearby frames (either motioncentric or to the entire frame) [5, 8, 15, 6, 21].”
Structure-Preserving Binary Representations for RGB-D Action Recognition (2016, methodology): “Fathi and Mori [30] developed a method to extract mid-level motion features by using the low-level optical ﬂow for action recognition.”
Recognizing human actions by attributes (2011, unknown): “Some researchers have proposed to represent action by intermediate semantic features [19, 20, 9], which are conceptually similar to our data-driven attributes.”
A Review of Human Activity Recognition Methods (2015, methodology): “The approach of Fathi and Mori (2008) was based on mid-level motion features, which are also constructed directly from optical flow features.”
Learning a hierarchy of discriminative space-time neighborhood features for human action recognition (2010, background): “To identify informative local video descriptors, boosting [12, 6], PageRank [18], and item-set mining [23, 9] have all been explored.”
Learning Sparse Representations for Human Action Recognition (2012, methodology): “Fathi & Mori [34] 51.”