DIBS: Enhancing Dense Video Captioning with Unlabeled Videos via Pseudo Boundary Enrichment and Online Refinement.
Hao WuHuabin LiuYu QiaoXiao SunPublished in: CoRR (2024)
Keyphrases
- online video
- video content
- video frames
- video sequences
- video data
- video database
- video analysis
- video clips
- video surveillance
- key frames
- video segments
- video editing
- video streams
- event recognition
- input video
- temporal coherence
- video event
- video representation
- natural language descriptions
- youtube videos
- human actions
- online learning
- high definition
- real time
- human activities
- video images
- video indexing
- video classification
- video shots
- dynamic scenes
- video retrieval
- instructional videos
- surveillance videos
- video browsing
- motion features
- space time
- video dataset
- video material
- video sharing
- spatiotemporal features
- semantic concept detection
- content based copy detection
- lecture videos
- video scene
- video annotation
- user generated
- video search
- successive frames
- motion estimation
- video quality assessment
- video objects
- stationary camera
- visual analysis
- stereoscopic video
- sports video
- web videos
- news video
- spatio temporal
- tv series
- moving objects
- multimedia