Combining CNN and LSTM for activity of daily living recognition with a 3D matrix skeleton representation.
Giovanni ErcolanoSilvia RossiPublished in: Intell. Serv. Robotics (2021)
Keyphrases
- daily living
- daily activities
- recognition accuracy
- convolutional neural network
- recognition rate
- activity recognition
- binary images
- human activities
- object recognition
- matrix representation
- action recognition
- human activity recognition
- feature extraction
- supervised learning
- machine learning
- singular value decomposition
- semi supervised
- delaunay triangulation
- pattern recognition