Fully-Coupled Two-Stream Spatiotemporal Networks for Extremely Low Resolution Action Recognition.
Mingze XuAidean SharghiXin ChenDavid J. CrandallPublished in: WACV (2018)
Keyphrases
- action recognition
- low resolution
- high resolution
- super resolution
- low resolution images
- human actions
- activity recognition
- bag of words
- computer vision
- high quality
- higher resolution
- image super resolution
- face images
- human activities
- data streams
- space time
- motion estimation
- image patches
- depth map
- moving objects
- training set
- active appearance models
- depth images