Regularizing Long Short Term Memory with 3D Human-Skeleton Sequences for Action Recognition

By Behrooz Mahasseni, Sinisa Todorovic. 2016

166 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Model Architectures

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

In Model Architectures

Citations in the corpus

RGB-D-based Human Motion Recognition with Deep Learning: A Survey (2017, methodology): “For RGB and skeleton fusion, Mahassein and Todorovic [150] presented a regularization of LSTM learning where the output of another encoder LSTM (eLSTM) grounded on 3D human-skeleton training data is used as the regularization.”
Skeleton-Based Action Recognition Using Spatio-Temporal LSTM Network with Trust Gates (2017, methodology): “Mahasseni and Todorovic [61] proposed to use LSTM to encode a skeleton sequence as a feature vector.”
Chained Multi-stream Networks Exploiting Pose, Motion, and Appearance for Action Classification and Detection (2017, neutral): “Several works employed deep architectures for classification [26, 41, 45, 34].”
Learning and Refining of Privileged Information-Based RNNs for Action Recognition from Depth Sequences (2017, methodology): “More recently, [30] present a regularized RNNs with additional information for RGB video sequences.”
Lattice Long Short-Term Memory for Human Action Recognition (2017, methodology): “[22] proposes to train an LSTM that is regularized using the output of another encoder LSTM (RLSTM-g3) grounded on 3D human-skeleton training data, while [2] proposes a two-step learning method in which 3D features are extracted using CNNs, and then an RNN is trained to classify each sequence by con”
Human Action Recognition From Various Data Modalities: A Review (2020, methodology): “and Todorovic [24] utilized a CNN+LSTM network to perform classification based on RGB videos, and an LSTM model trained on skeleton data to act as a regularizer by”
On Geometric Features for Skeleton-Based Action Recognition Using Multilayer LSTM Networks (2017, unknown): “, video classification [6, 33, 17, 24] and image description [6, 30].”
Fusing Geometric Features for Skeleton-Based Action Recognition Using Multilayer LSTM Networks (2018, unknown): “, video classification [1], [18], [36], [37] and image description [1], [38].”
Leveraging Structural Context Models and Ranking Score Fusion for Human Interaction Prediction (2016, background): “They are capable of detecting salient keywords in sequential data, such as speech, sentences and actions [12]–[15].”
A Survey on Deep Learning Based Approaches for Action and Gesture Recognition in Image Sequences (2017, methodology): “[42] uses an LSTM to model each individual’s actions and a second-level LSTM aggregates the outputs of individual LSTMs.”
Multimodal Machine Learning: A Survey and Taxonomy (2017, background): “Non-parallel Transfer learning Visual classification [64] Action recognition [134] Concept grounding Metaphor class.”
On Human Motion Prediction Using Recurrent Neural Networks (2017, background): “Although their purpose (action classification from a sequence of poses) is different from ours, this field contains interesting insights for motion prediction, such as the importance of a mathematically sound orientation representation [46,47] or how learned, compact motion representations improve a”
Spatiotemporal Multiplier Networks for Video Action Recognition (2017, unknown): “[23, 25, 27, 40]).”