Unsupervised Learning of Human Action Categories Using Spatial-Temporal Words

By Juan Carlos Niebles, Hongchen Wang, Hongcheng Wang, Li Fei-Fei. 2008

1,937 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

Unsupervised Learning from Narrated Instruction Videos (2015, background): “More recently, weakly supervised learning of actions in video using video scripts or event order has been addressed in [3, 4, 5, 9, 16].”
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, 21, 30].”
The Kinetics Human Action Video Dataset (2017, background): “This is in the spirit of bag of words image modeling approaches [16, 17, 24], but while convenient in practice, it has the issue of entirely ignoring temporal structure (e.g. models can’t potentially distinguish opening from closing a door).”
Long-Term Temporal Convolutions for Action Recognition (2016, neutral): “Such a view finds support both in psychology [1] and in computer vision approaches to action recognition in video [2], [3], [4], [5].”
Video In Sentences Out (2012, methodology): “Extracting the information needed to render these linguistic entities requires an approach to event recognition that recovers object tracks, the track-to-role assignments, and changing body posture.”
Human Action Recognition Using Factorized Spatio-Temporal Convolutional Networks (2015, background): “Note that our proposed scheme is different from the bag of video words [22], which extracts spatio-temporal features from the video cuboid.”
Connectionist Temporal Modeling for Weakly Supervised Action Labeling (2016, methodology): “Our use of visual similarities in the training is related to unsupervised video parsing [28,29,45], where frames are grouped into segments based on visual or semantic cues.”
A Biologically Inspired System for Action Recognition (2007, background): “The other common class of approaches is based on the processing of spatio-temporal features, either global as in the case of low-resolution videos [33, 6, 2] or local for higher resolution images [24, 5, 7, 18].”
Hough Forests for Object Detection, Tracking, and Action Recognition (2011, background): “Based on the success of such systems, recent works in tracking and in action recognition have extended many successful ideas into the spatio-temporal domain [10], [11], [12], [13], [14], [15].”
Slow Feature Analysis for Human Action Recognition (2012, methodology): “In this paper, we introduce the SFA framework to the problem of human action recognition by incorporating the discriminative information with SFA learning and considering the spatial relationship of body parts.”
An efficient part-based approach to action recognition from RGB-D video with BoW-pyramid representation (2013, methodology): “[17] followed the approach in [14] to detect interest points, and they used two models, i.”
Modeling Complex Temporal Composition of Actionlets for Activity Prediction (2012, background): “Then, actionlets categories are learned in an unsupervised way from histogram of spatial-temporal words [17].”
Discriminative subvolume search for efficient action detection (2009, methodology): “Contemporary methods for action categorization use local spatio-temporal features to characterize the video and perform classification over the set of local features [13] [12] [17] [19] To improve the classification performance, both shape and motion information are applied for action categorization”
Learning the semantics of object–action relations by observation (2011, background): “…much attention in recent years and many contributions exist, but are often unrelated to manipulation recognition (Calinon and Billard 2004, 2005; Hakeem and Shah 2005; Maurer et al. 2005; Calinon and Billard 2007; Laptev and Perez 2007; Niebles et al. 2008; Dee et al. 2009; Gilbert et al. 2011).”
Pose primitive based human action recognition in videos or still images (2008, background): “In [16] human action-categories are represented and learned using spacetime interest points, in [15] a hierarchical bag of features approach is applied to image sequences.”
Learning a hierarchy of discriminative space-time neighborhood features for human action recognition (2010, methodology): “Almost all reported numbers use this setup, though [9] also report leave-one-out accuracy, while [3, 5, 21] exclusively use leave-one-out.”
SIFT-Bag kernel for video event analysis (2008, methodology): “Categories and Subject Descriptors I.4.9 [Computing Methodologies]: Image Processing and Computer Vision Applications General Terms Algorithm, Performance, Experimentations Keywords SIFT-Bag, Kernel Design, Within-Class Covariation Normalization, Video Event Recognition 1.”