Recur, Attend or Convolve? Frame Dependency Modeling Matters for Cross-Domain Robustness in Action Recognition.
Sofia BrooméErnest PokropekBoyu LiHedvig KjellströmPublished in: CoRR (2021)
Keyphrases
- action recognition
- cross domain
- human actions
- bag of words
- action classification
- activity recognition
- computer vision
- transfer learning
- body parts
- domain adaptation
- knowledge transfer
- recognition of human actions
- recognizing human actions
- human activities
- human pose
- recognizing actions
- view invariant
- human object interactions
- active learning
- action detection
- space time interest points
- text classification
- feature extraction