Marginalised Stacked Denoising Autoencoders for Robust Representation of Real-Time Multi-View Action Recognition.
Feng GuFrancisco Flórez-RevueltaDorothy MonekossoPaolo RemagninoPublished in: Sensors (2015)
Keyphrases
- view invariant
- multi view
- denoising
- action recognition
- single view
- multiple views
- human actions
- multiple viewpoints
- depth map
- multiple cameras
- image processing
- three dimensional
- natural images
- view synthesis
- recognizing human actions
- d objects
- semi supervised
- surface reconstruction
- computer vision
- motion capture
- recognition of human actions
- action recognition in videos
- human motion
- activity recognition
- range images
- visual hull
- super resolution
- machine learning