Deep Optical Flow Feature Fusion Based on 3D Convolutional Networks for Video Action Recognition.
Tongwei LuShihui AiYongyuan JiangYudian XiongFeng MinPublished in: SmartWorld/SCALCOM/UIC/ATC/CBDCom/IOP/SCI (2018)
Keyphrases
- action recognition
- feature fusion
- motion features
- human actions
- optical flow
- action classification
- action detection
- computer vision
- recognizing human actions
- recognition of human actions
- activity recognition
- feature extraction
- space time interest points
- bag of words
- multiple features
- video data
- human activities
- video sequences
- least squares
- image sequences
- data fusion
- multi sensor
- subspace learning
- moving objects
- atomic actions
- motion history images
- viewpoint
- high level
- motion estimation
- spatio temporal
- visual words
- machine learning