Human Action Recognition Based on MOCAP Information Using Convolution Neural Networks

55 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

RGB-D-based action recognition datasets: A survey (2016, neutral): “ne Frame, OnFr: Online Frame. Dataset State-of-the-art Methods Acc.(%) Data Protocol Berkeley MHAD [42] 1. Hierarchy of LDSs(28 joints used) [93] 2. HBRNN-L(35 joints used) [94] 3. CNN(3 joints used) [95] 4. Feature-Level-Fusion+SRC (Kinect+Acc1&Acc4) [96] 5. HACK [97] 1. 100 2. 100 3. 98.28 4. ”
Graph Based Convolutional Neural Network (2016, methodology): “Methods of handling such an irregular space as an input include using standard neural networks, embedding the feature space onto a grid to allow convolution [8], identifying local patches on the irregular manifold to perform geodesic convolutions [14], or graph signal processing based convolutions o”
3D skeleton-based human action classification: A survey (2016, methodology): “In [107], only the joints corresponding to the right-hand, the left-hand and the pelvis are used to extract features relevant for the actions to recognize.”