Object-centric Video Prediction without Annotation.
Karl SchmeckpeperGeorgios GeorgakisKostas DaniilidisPublished in: CoRR (2021)
Keyphrases
- weakly labeled
- prediction accuracy
- video annotation
- video data
- d objects
- video streams
- multimedia
- real time
- prediction error
- video sequences
- video analysis
- prediction model
- successive frames
- object detection and tracking
- object motion
- object tracking
- active learning
- video material
- metadata
- complex objects
- stationary camera
- automatic annotation
- object recognition
- manual annotation
- temporal continuity
- combining information from multiple
- semantic labels
- textual descriptions
- video database
- video clips
- video retrieval
- space time
- computer vision