Spatio-Temporal FAST 3D Convolutions for Human Action Recognition

Computer Science

• Convolutional Neural Network 100%
• human action recognition 100%
• Vertical Direction 100%
• Horizontal Direction 100%
• Temporal Characteristic 50%
• Action Recognition 50%
• Convolution Operation 50%
• Residual Neural Network 50%
• Memory Requirement 50%
• DenseNet 50%
• Vertical Motion 50%

Keyphrases

• Spatiotemporal 100%
• Human Action Recognition 100%
• 3D Convolution 100%
• Convolution Block 57%
• Temporal Convolutional Network 28%
• CNN Architecture 28%
• Video Input 28%
• Temporal Features 14%
• Vertical Motion 14%
• Human Action 14%
• Horizontal Motion 14%
• Memory Requirements 14%
• Increased Performance 14%
• Natural Decomposition 14%
• Spatial Convolution 14%
• HMDB51 14%
• UCF-101 14%
• Two-stream Network 14%
• Convolution Operation 14%
• ResNet Architecture 14%
• DenseNet121 14%
• Action Recognition Dataset 14%
• Validation Loss 14%
• Deep Network 14%

Mathematics

• Convolution 100%
• Human Action 100%
• Vertical Direction 15%
• Convolutional Neural Network 15%