Activity recognition and abnormality detection with the switching hidden semi-Markov model

By Thi V. Duong, Hung Hai Bui, Dinh Q. Phung, Svetha Venkatesh. Computer Vision and Pattern Recognition, 2005

613 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Surgery

View Paper: https://www.semanticscholar.org/paper/4ad862b9152c81ad13f9c0ed94ee302c63b8a64c

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

In Surgery

Citations in the corpus

A Scalable Approach to Activity Recognition based on Object Use (2007, methodology): “Temporal constraints on actions are addressed either through the use of probabilistic models such as HMM’s [4, 16] or SCFG’s [9], or through explicit temporal correlation methods [21, 26].”
Human Activity Recognition Using Recurrent Neural Networks (2017, methodology): “Patterns corresponding to the activities are detected using sensors such as accelerometers, gyroscopes or passive infrared sensors, etc. , either using feature extraction on sliding window followed by classiﬁcation [4] or with Hidden Markov Modeling (HMM) [5].”
Simultaneous Plan Recognition and Monitoring (SPRAM) for Robot Assistants (2013, background): “the inclusion of durations [6] or hierarchies [11, 4].”
Learning latent spatio-temporal compositional model for human action recognition (2013, background): “A lot of works were proposed to build temporal structure models [9, 25, 35, 4] based on discriminative and interesting motion segments of the video.”
A system for learning statistical motion patterns (2006, background): “Duong et al. [11] apply the Switching Hidden Semi-Markov Model to learn and recognize human behaviors and detect anomalies.”
Accurate activity recognition in a home setting (2008, methodology): “Finally, hierarchical hidden Markov models were used to perform activity recognition from video [5].”
A Review on Video-Based Human Activity Recognition (2013, methodology): “In [1], the kitchen is quantized into 28 square cells of 1 m(2) each and the position of the human is captured by four cameras mounted at the ceiling corners, and the tracking system returns the list of cells visited by the person as the moving trajectory path.”
Human Action Recognition and Prediction: A Survey (2018, methodology): “2 Sequential Approaches: This line of work captures temporal evolution of appearance or pose using sequential state models such as hidden Markov models (HMMs) [108], [109], [110], conditional random fields (CRFs) [83], [84], [111], [112] and structured support vector machine (SSVM) [13], [54], [85],”
Video Behavior Profiling for Anomaly Detection (2008, background): “Much work on abnormal behavior detection 1 took a supervised learning approach [16], [14], [8], [5], [3] based on the assumption that there exist well-defined and known a priori behavior classes (both normal and abnormal).”
Advanced internet of things for personalised healthcare systems: A survey (2017, methodology): “Sub-Categories Algorithms References Data-driven approaches Supervised learning ANNs [23],[90], [150] HMMs [93]–[100] SVM [34], [96], [42], [111], [123], [124] Decision tree [36] , [126]–[130] Semi-supervised learning Co-training [133], [38], [109], [110] Unsupervised learning Expectation maximum…”
Learning latent temporal structure for complex event detection (2012, methodology): “Our model for videos is the conditional variant of the variable-duration hidden Markov model (HMM), also referred to as an explicit-duration HMM or a hidden semiMarkov model [2, 14].”
Sensor-Based Abnormal Human-Activity Detection (2008, background): “These approaches range from the computer vision area [6], [27], where variations of Markov models are designed to detect the out-of-norm behavior of users, to the data mining area [1], [12], where similarity or model-based approaches are employed to detect outliers.”
Poselet Key-Framing: A Model for Human Activity Recognition (2013, methodology): “Sequential models: Hidden semi-Markov Models (HSMM) [10] , CRFs [31], and finite-state-machines [13] have been used to model the temporal evolution of human activities.”
Understanding Video Events: A Survey of Methods for Automatic Interpretation of Semantic Occurrences in Video (2009, methodology): “Similarly, other works, emphasizing event modeling, choose straightforward abstraction schemes [17], [24], [52], [86].”
Modelling Crowd Scenes for Event Detection (2006, background): “The application domains for these techniques usually involve simple environments such as offices [8], kitchens [3] , cargo bays [6] and loading docks [5] such that activity recognition is focused upon modeling the actions and interactions of small groups of people/objects.”