JEDI: Joint Expert Distillation in a Semi-Supervised Multi-Dataset Student-Teacher Scenario for Video Action Recognition.
Lucian BicsiBogdan AlexeRadu Tudor IonescuMarius LeordeanuPublished in: CoRR (2023)
Keyphrases
- action recognition
- human actions
- semi supervised
- video dataset
- action classification
- space time interest points
- spatial temporal
- student teachers
- bag of words
- activity recognition
- action detection
- recognition of human actions
- visual data
- human activities
- computer vision
- video sequences
- body parts
- recognizing human actions
- recognizing actions
- space time
- human motion
- human pose
- spatio temporal
- multi view
- motion features
- video shots
- labeled data
- static images
- active learning
- video data
- action recognition in videos
- e learning
- video frames