/action-analysis-review

This review contains some prevailing methods for action recognition.

action recognition review

(1). Datasets

(2). Methods

  • Two-stream based
    • Two-Stream Convolutional Networks for Action Recognition in Videos paper
    • Temporal Segment Networks: Towards Good Practices for Deep Action Recognition paper code
  • 3D convolution based
    • C3D: Generic Features for Video Analysis paper code
    • Temporal 3D ConvNets: New Architecture and Transfer Learning for Video Classification paper code
    • A Closer Look at Spatiotemporal Convolutions for Action Recognition paper code
    • Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset paper code
    • Non-local Neural Networks paper code
  • Skeleton based
    • Co-occurrence Feature Learning for Skeleton based Action Recognition using Regularized Deep LSTM Networks paper
    • An End-to-End Spatio-Temporal Attention Model for Human Action Recognition from Skeleton Data paper project
    • View Adaptive Neural Networks for High Performance Skeleton-based Human Action Recognition paper
    • Co-occurrence Feature Learning from Skeleton Data for Action Recognition and Detection with Hierarchical Aggregation paper
    • Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition paper code
  • Others
    • ActionVLAD: Learning spatio-temporal aggregation for action classification paper code

    • Appearance-and-Relation Networks for Video Classification paper code

    • ECO: Efficient Convolutional Network for Online Video Understanding paper code