Action Recognition From Depth Maps Using Deep Convolutional Neural Networks

By Pichao Wang, Wanqing Li, Zhimin Gao, Jing Zhang, Chang-Fu Tang, Philip O. Ogunbona. 2016

293 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Biomedical Engineering, Computer Vision and Pattern Recognition, 3D Vision & Pose

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 Biomedical Engineering

In Computer Vision and Pattern Recognition

In 3D Vision & Pose

Citations in the corpus

NTU RGB+D: A Large Scale Dataset for 3D Human Activity Analysis (2016, background): “Although some attempts have been done [9, 42], they suffered from overﬁtting and had to scale down the size of learning parameters; as a result, they clearly need many more samples to generalize and perform better on testing data.”
NTU RGB+D 120: A Large-Scale Benchmark for 3D Human Activity Understanding (2019, background): “Though several attempts have been done [9], [10], they mostly suffer from over-ﬁtting and have to scale down the size of their learning models.”
Deep Multimodal Feature Analysis for Action Recognition in RGB+D Videos (2016, background): “The work of [77] applied convolutional networks for learning action classes on depth maps.”
Learning and Refining of Privileged Information-Based RNNs for Action Recognition from Depth Sequences (2017, background): “Action recognition from depth sequences [57, 34, 29, 44, 49] has attracted significant interest recently due to the emergence of low-cost depth sensors.”
Spatially and Temporally Structured Global to Local Aggregation of Dynamic Depth Information for Action Recognition (2018, methodology): “The last category is to represent video with one or multiple compact images and adopt available trained ConvNet architectures for fine-tuning [14], [15], [17], [29], [42]–[44].”
Joint Geometrical and Statistical Alignment for Visual Domain Adaptation (2017, methodology): “There are 10 common actions between MSRAction3DExt and UTD-MHAD: wave, hand catch, right arm high throw, draw x, draw circle, two hand front clap, jogging, tennis swing, tennis serve, and pickup and throw.”
A Review on Human Activity Recognition Using Vision-Based Method (2017, background): “Wang et al. [114] developed three strategies to leverage ConvNets on small training datasets.”
Robust 3D Hand Pose Estimation in Single Depth Images: From Single-View CNN to Multi-View CNNs (2016, background): “Multi-view CNNs have shown superior performance in 3D object recognition and retrieval [32], [33], as well as human action recognition [34].”
Human Action Recognition From Various Data Modalities: A Review (2020, methodology): “(DMM) features [220], a deep learning framework utilizing weighted hierarchical DMMs was proposed in [221].”
On Geometric Features for Skeleton-Based Action Recognition Using Multilayer LSTM Networks (2017, unknown): “Action recognition aims to identify human actions from input sensor streams, where RGB [7, 36], depth [31, 28] and skeleton [8, 37] are three common types of input.”
Depth Pooling Based Large-Scale 3-D Action Recognition With Convolutional Neural Networks (2018, methodology): “4) Computational cost: we compared the computational cost of the proposed method in the test phase without any code optimization with the computational cost of the RGB+D-based methods MFSK+DeepID [58], SFAM [55], depth-based method WHDMM [40] on ChaLearn LAP IsoGD dataset.”
Fusing Geometric Features for Skeleton-Based Action Recognition Using Multilayer LSTM Networks (2018, unknown): “2802648 [2], depth [3], [4] and skeleton [5], [6].”
Large-scale Isolated Gesture Recognition using Convolutional Neural Networks (2016, background): “Which one among these and other possibilities would result in the best representation in the context of gesture recognition is not well understood.”
Spatio-Temporal Graph Convolution for Skeleton Based Action Recognition (2018, methodology): “We follow the standard experimental settings to perform leave-one-subject-out cross-validation (Wang et al. 2016a).”