Skeletal Quads: Human Action Recognition Using Joint Quadruples

By Gurkirt Singh, Radu Patrice Horaud. 2014

275 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

In Computer Vision and Pattern Recognition

Citations in the corpus

NTU RGB+D: A Large Scale Dataset for 3D Human Activity Analysis (2016, methodology): “…Accuracy HOG 2 [24] 32.24% 22.27% Super Normal Vector [44] 31.82% 13.61% HON4D [25] 30.56% 7.26% Lie Group [37] 50.08% 52.76% Skeletal Quads [10] 38.62% 41.36% FTP Dynamic Skeletons [15] 60.23% 65.22% HBRNN-L [9] 59.07% 63.97% This shows they have their model was prone to overﬁtting due to…”
NTU RGB+D 120: A Large-Scale Benchmark for 3D Human Activity Understanding (2019, background): “Evangelidis et al. [53] divided the body into part-based joint quadruples, and encoded the conﬁguration of each part with a succinct 6D feature vector, so called skeletal quads.”
A self-organizing neural network architecture for learning human-object interactions (2017, methodology): “We extract the skeletal quad features [49], which are invariant with respect to location, viewpoint as well as body-orientation.”
Deep Multimodal Feature Analysis for Action Recognition in RGB+D Videos (2016, background): “Recent single modality action recognition methods on depth signals can be divided into two major groups: depth map analysis methods [35, 41, 45, 79] and skeleton based methods [36, 65, 11, 73, 74].”
Space-Time Representation of People Based on 3D Skeletal Data: A Review (2016, background): “[193] Skeletal Quad Conc Lowlv Hand X X X Azary and Savakis [194] Grassmann Manifolds Conc Manif Hand X X X X Barnachon et al.”
Glimpse Clouds: Human Activity Recognition from Unstructured Feature Points (2018, neutral): “5 Skeleton Quads [13] - 38.”
When Kernel Methods Meet Feature Learning: Log-Covariance Network for Action Recognition From Skeletal Data (2017, methodology): “Among others, we can mention histogram based representations to perform temporal pooling [42, 41, 6], extraction of local spatio-temporal features from the data [4, 30], also applying bag-of-words or Fisher vector approaches to aggregate the raw joint representation in a unique action descriptor [6,”
Skeleton-Based Action Recognition Using Spatio-Temporal LSTM Network with Trust Gates (2017, background): “[10] propose to learn a GMM over the Fisher kernel representation of the skeletal quads feature.”
Modeling Temporal Dynamics and Spatial Configurations of Actions Using Two-Stream Recurrent Neural Networks (2017, background): “, 3D skeletons representation in a Lie group [44], Fisher vector encoding of skeletal quads [10] and FTP dynamic [18].”
View Adaptive Recurrent Neural Networks for High Performance Human Action Recognition from Skeleton Data (2017, neutral): “CS CV Skeleton Quads [6] 38.6 41.4 Lie Group [39] 50.1 52.8 Dynamic Skeletons [13] 60.2 65.2 HBRNN-L [5] 59.1 64.0 Part-aware LSTM [31] 62.9 70.3 ST-LSTM (Tree Traversal) + Trust Gate [24] 69.2 77.7 STA-LSTM [35] 73.4 81.2 VA-LSTM 79.4 87.6 Table 2: Comparisons on the SBU dataset in accuracy (%).”
A New Representation of Skeleton Sequences for 3D Action Recognition (2017, neutral): “Accuracy Cross Subject Cross View Lie Group [28] 50.1% 52.8% Skeletal Quads [5] 38.6% 41.4% Dynamic Skeletons [11] 60.2% 65.2% Hierarchical RNN [4] 59.1% 64.0% Deep RNN [24] 59.3% 64.1% Deep LSTM [24] 60.7% 67.3% Part-aware LSTM [24] 62.9% 70.3% ST-LSTM [16] 65.2% 76.1% ST-LSTM + Trust Gate [16] 69…”
Unsupervised Learning of Long-Term Motion Dynamics for Videos (2017, methodology): “…into training and testing groups Methods Depth HOG [17] 32.24 Super Normal Vector [42] 31.82 HON4D [18] 30.56 Lie Group [32] 50.08 Skeletal Quads [3] 38.62 FTP Dynamic Skeletons [4] 60.23 HBRNN-L [2] 59.07 2 Layer P-LSTM [25] 62.93 Shufﬂe and Learn [15] 47.5 Our method (Unsupervised training)…”
Spatio-Temporal LSTM with Trust Gates for 3D Human Action Recognition (2016, background): “Cross subject Cross view Lie Group [7] 50.1% 52.8% Skeletal Quads [6] 38.6% 41.4% Dynamic Skeletons [61] 60.2% 65.2% HBRNN [30] 59.1% 64.0% Part-aware LSTM [32] 62.9% 70.3% Deep RNN [32] 56.3% 64.1% Deep LSTM [32] 60.7% The results are shown in Table 1.”
Action Recognition Based on Joint Trajectory Maps Using Convolutional Neural Networks (2016, methodology): “VOL. X, NO. X, X 2016 (WILL BE INSERTED BY THE EDITOR) 2 Scores Scores Scores Multiply Score Fusion Accuracy Front Top Side Front Channel Top Channel [59], geometric descriptor [33], [34], [39]–[41], [60] or keySide Channel Skeleton Sequences ConvNet ConvNet ConvNet Rotation Color Encoding Color Enc”
Fusing shape and motion matrices for view invariant action recognition using 3D skeletons (2017, methodology): “[15] presented a local skeleton descriptor named skeletal quads, further Fisher kernel representation was applied to describe the skeletal quads, and linear SVMwas used to classify action.”
Multitype Activity Recognition in Robot-Centric Scenarios (2015, unknown): “[34] - 89.”