FeatFSDA: Towards Few-shot Domain Adaptation for Video-based Activity Recognition.
Kunyu PengDi WenDavid SchneiderJiaming ZhangKailun YangM. Saquib SarfrazRainer StiefelhagenAlina RoitbergPublished in: CoRR (2023)
Keyphrases
- domain adaptation
- activity recognition
- transfer learning
- human activity recognition
- cross domain
- human activities
- target domain
- action recognition
- multiple sources
- video sequences
- labeled data
- semi supervised
- sentiment classification
- semi supervised learning
- video shots
- video content
- video data
- key frames
- unlabeled data
- knowledge transfer
- document classification
- accelerometer data
- visual features
- computer vision
- learning algorithm
- test data
- text classification
- text mining
- collaborative filtering
- active learning
- natural language
- co training
- similarity measure
- face recognition
- machine learning