Recognition and Segmentation of 3-D Human Action Using HMM and Multi-class AdaBoost

By Fengjun Lv, Ramkant Nevatia. 2006

418 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Human-Computer Interaction

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

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

In Human-Computer Interaction

1. Wanqing Li
2. Tae-Kyun Kim
3. Hossein Rahmani
4. Jun Liu
5. G. Wang
6. Liang Wang
7. Yun Raymond Fu
8. Amir Shahroudy

Citations in the corpus

• Space-Time Representation of People Based on 3D Skeletal Data: A Review (2016, background): “Lv and Nevatia [203] HMM/Adaboost Conc Lowlv Hand X X X Pons-Moll et al.”
• Modeling Temporal Dynamics and Spatial Configurations of Actions Using Two-Stream Recurrent Neural Networks (2017, background): “Joint based approaches consider the human skeleton as a set of points and use various positions based features such as joint positions [20, 31] and pairwise relative joint positions [46, 51] to characterize actions.”
• Deep Learning on Lie Groups for Skeleton-Based Action Recognition (2016, background): “Due to the development of depth sensors, 3D human activity analysis [26, 44, 22, 42, 40, 3, 41, 36, 43, 25, 34, 17] has attracted more interest than ever before.”
• Ongoing human action recognition with motion capture (2014, methodology): “Except for [19], where depth maps were used as inputs, the other four methods [21,23,25,40] as well as ours use skeleton inputs, extracted from video streams.”
• Kernelized covariance for action recognition (2016, methodology): “For instance, a hidden Markov model is used in [4] to produce weak classifiers which are enhanced by AdaBoost.”
• Hierarchical Aligned Cluster Analysis for Temporal Clustering of Human Motion (2013, background): “Unsupervised techniques to learn motion primitives from training data have recently attracted the interest of many scientists in computer vision [4], [5], [6], [7], [8], [9], [10], [11], [12] and computer graphics [13], [14], [15], [16], [17], [18], [19], [20].”
• The Moving Pose: An Efficient 3D Kinematics Descriptor for Low-Latency Action Recognition and Detection (2013, neutral): “Method Accuracy(”
• Learning Clip Representations for Skeleton-Based 3D Action Recognition (2018, methodology): “(HMMs) [19], [20], Conditional Random Fields (CRFs) [21] and Temporal Pyramids (TPs) [7], [22] have been used to”
• Skeleton based action recognition with convolutional neural network (2015, methodology): “Alternatively, some time series models, mainly Dynamic TimeWarpings (DTWs) [19] and Hidden Markov Models (HMMs) [11, 21, 23], are applied to model the global temporal evolution, yet it is very difficult to obtain the temporal aligned sequences and the emission distributions of HMMs.”
• Spatiotemporal representation of 3D skeleton joints-based action recognition using modified spherical harmonics (2016, methodology): “In (Lv and Nevatia, 2006), the 3D coordinates of the joints were used for a skeleton representation of the human body.”
• Transition Forests: Learning Discriminative Temporal Transitions for Action Recognition and Detection (2016, methodology): “[12] derive local features from individual and multiple joints, while [22] represents poses as 3D histograms of joints for training a HMM.”
• Enhanced Sequence Matching for Action Recognition from 3D Skeletal Data (2014, neutral): “Hidden Markov Model (HMM) [16–18, 8] and Conditional Random Field (CRF) [19–21] are popular models for this approach.”
• Action recognition using ensemble weighted multi-instance learning (2014, methodology): “Lv and Nevatia [12] employed a hidden markov model (HMM) to represent the transition probability for pre-defined 3D joint”
• Action Recognition Using Rate-Invariant Analysis of Skeletal Shape Trajectories (2016, methodology): “Some of these datasets are also used in the experimental studies described later.”
• HON4D: Histogram of Oriented 4D Normals for Activity Recognition from Depth Sequences (2013, methodology): “On the other hand, parallel to the approaches developed for temporal modelling of human actions in color videos such as [15, 2], Lv and Nevatia in [14] employ a hidden Markov model (HMM) to represent the transition probability for pre-defined 3D joint positions.”