Action Recognition by Learning Deep Multi-Granular Spatio-Temporal Video Representation

129 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.

Video Representation Learning Using Discriminative Pooling (2018, background): “This problem may be tackled using recurrent models [1, 8, 9, 25, 45, 60], however such models are usually harder to train [32].”
Skeleton-Based Action Recognition Using Spatio-Temporal LSTM Network with Trust Gates (2017, neutral): “[25], [26], [27], [28], [29], [30], and RGB-based activity recognition [31], [32], [33], [34], [35], [36].”
Chained Multi-stream Networks Exploiting Pose, Motion, and Appearance for Action Classification and Detection (2017, methodology): “The flow and magnitude values in the image are multiplied by 16 and quantized into the [0,255] interval [20, 31, 46, 47].”
VidTr: Video Transformer Without Convolutions (2021, methodology): “The LSTM [25] was later proposed to model the image feature based on ConvNet features [30, 51, 63].”
Ordered Pooling of Optical Flow Sequences for Action Recognition (2017, methodology): “We also note that there have been several other deep learning models devised for action modeling such as using 3D convolutional filters, recurrent neural networks [3, 6] and long-short term memory networks [5, 16, 31].”
Learning Spatio-Temporal Representation with Pseudo-3D Residual Networks (2017, background): “This architecture is further extended by exploiting multi-granular structure [17, 18, 21], convolutional fusion [6], key-volume mining [39] and temporal segment networks [36] for video representation learning.”