Privacy-Preserving Action Recognition via Motion Difference Quantization.
Sudhakar KumawatHajime NagaharaPublished in: ECCV (13) (2022)
Keyphrases
- privacy preserving
- action recognition
- recognizing human actions
- recognition of human actions
- human movements
- static images
- action recognition in videos
- motion features
- privacy preserving data mining
- human actions
- action primitives
- privacy preservation
- view invariant
- activity recognition
- motion capture data
- vertically partitioned data
- motion history images
- bag of words
- action classification
- computer vision
- body parts
- horizontally partitioned data
- privacy sensitive
- privacy concerns
- privacy preserving association rule mining
- motion estimation
- sensitive data
- data privacy
- private information
- sensitive information
- image sequences
- scalar product
- optical flow
- motion model
- human activities
- human motion
- secure multiparty computation
- motion capture
- bag of features
- space time
- privacy protection
- robust tracking
- human activity recognition
- decision trees
- machine learning