Prediction of Human Activity by Discovering Temporal Sequence Patterns

By Kang Li, Yun Raymond Fu. 2014

229 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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

1. Jun Liu
2. G. Wang
3. Amir Shahroudy
4. Alex ChiChung Kot
5. Yun Raymond Fu
6. Ling-yu Duan
7. Li Fei-Fei
8. Yu Kong

In Computer Vision and Pattern Recognition

1. G. Wang
2. Jun Liu
3. Luc Van Gool
4. Xiaogang Wang
5. Li Fei-Fei
6. Ming-Hsuan Yang
7. Dacheng Tao
8. Yi Yang

Citations in the corpus

• Skeleton-Based Online Action Prediction Using Scale Selection Network (2019, background): “Predicting (recognizing) an action before it gets fully performed has attracted a lot of research attention recently [2, 3, 50–54].”
• Joint Prediction of Activity Labels and Starting Times in Untrimmed Videos (2017, methodology): “There have been a few works which predict the label of the future unobserved activity such as approaches using semantic scene labeling [19], Probabilistic Suffix Tree (PST) [23], augmentedHidden Conditional Random Field (a-HCRF) [44], Markov Random Field (MRF) [3], kernel-based reinforcement learnin”
• Social LSTM: Human Trajectory Prediction in Crowded Spaces (2016, methodology): “[63, 40] used a similar idea to identify the functional map of a scene.”
• Activity Auto-Completion: Predicting Human Activities from Partial Videos (2015, background): “In addition, Li and Fu [16] focus on solving long-duration complex activity prediction problem through sequential pattern mining, which is different from our goal.”
• Deep Sequential Context Networks for Action Prediction (2017, background): “Our work focuses on short-duration prediction such as “biking” and “diving”, while [17] focuses on long-duration compositional action prediction where an action can be further decomposed into semantic meaningful primitives.”
• Hierarchical recurrent neural network for skeleton based action recognition (2015, background): “Human skeleton based action recognition is generally considered as a time series problem [5, 17], in which the characteristics of body postures and their dynamics over time are extracted to represent a human action.”
• Predicting the Where and What of Actors and Actions through Online Action Localization (2016, background): “There have been recent efforts to predict activities by early recognition [18, 16, 26, 14].”
• First-Person Activity Forecasting with Online Inverse Reinforcement Learning (2016, background): “Other works predict actions several seconds into the future [3, 8, 11, 29].”
• Computer Vision and Image Understanding Computer Vision for Assistive Technologies (unknown): “Frameworks for long-duration complex activity prediction by discovering three key aspects of activity (Causality, Context-cue, and Predictability ) have been also proposed ( Li and Fu, 2014 ).”
• A Survey of Methods for Safe Human-Robot Interaction (2017, background): “While a majority of the previously mentioned works focus on shortterm prediction of brief actions, Li and Fu (2014) developed a framework for predicting actions of longer duration composed of constituent actions with complex temporal compositions.”
• Skeleton based action recognition with convolutional neural network (2015, background): “Temporal dynamics of postures over time can be modeled as a time series problem, which is crucial for sequencebased action recognition [4, 8, 9].”
• Real-Time RGB-D Activity Prediction by Soft Regression (2016, background): “Existing action/activity prediction model [4] requires manual labeling of a large amount of video segments, which however record mostly low level actions shared among activities, and thus it is too expensive and sometimes is difficult to label.”
• SSNet: Scale Selection Network for Online 3D Action Prediction (2018, background): “Recognizing (predicting) an action before it is fully performed has attracted a lot of attention recently [31, 50, 4, 71, 36, 27, 28].”
• Progressive Teacher-Student Learning for Early Action Prediction (2019, methodology): “Following the existing setting in [16, 24, 21, 32, 25, 26], we assume that each training video contains complete action execution and uniformly partition it into N shorter segments.”
• Learning hierarchical spatio-temporal pattern for human activity prediction (2016, methodology): “It should be noted that the same term actionlet has also been used in the recent work, which refers to action components based on a spatial segmentation [5] and component action units obtained from temporal decomposition [6].”
• Human Action Recognition and Prediction: A Survey (2018, background): “In their extension [143], object context is added to the prediction model, which enables the prediction of human-object interactions occurring in ac-”