A 3D-CNN and LSTM Based Multi-Task Learning Architecture for Action Recognition.
Xi OuyangShuangjie XuChaoyun ZhangPan ZhouYang YangGuanghui LiuXuelong LiPublished in: IEEE Access (2019)
Keyphrases
- action recognition
- multi task learning
- multi task
- human actions
- bag of words
- learning tasks
- max margin
- activity recognition
- gaussian processes
- computer vision
- transfer learning
- learning problems
- human detection
- body parts
- theoretical analysis
- recognizing human actions
- high order
- learning algorithm
- human activities
- human pose
- inductive learning
- pose estimation
- learning models
- active learning
- image segmentation