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