Fully-Coupled Two-Stream Spatiotemporal Networks for Extremely Low Resolution Action Recognition.
Mingze XuAidean SharghiXin ChenDavid J. CrandallPublished in: CoRR (2018)
Keyphrases
- action recognition
- low resolution
- high resolution
- super resolution
- human actions
- activity recognition
- low resolution images
- bag of words
- high quality
- face images
- image super resolution
- data streams
- higher resolution
- computer vision
- human activities
- image patches
- active appearance models
- space time
- depth map
- three dimensional
- depth images
- spatio temporal
- image reconstruction
- computational complexity
- image processing