COBE: Contextualized Object Embeddings from Narrated Instructional Video.
Gedas BertasiusLorenzo TorresaniPublished in: NeurIPS (2020)
Keyphrases
- multimedia
- video sequences
- video streams
- video content
- real time
- instructional design
- video frames
- video data
- successive frames
- d objects
- object motion
- student learning
- video scene
- video objects
- object tracking
- moving objects
- learning environment
- complex objects
- metadata
- object detection and tracking
- weakly labeled
- video database
- multiple objects
- dynamic scenes
- video retrieval
- learning outcomes
- low dimensional
- video clips
- video surveillance
- object model
- spatial relations
- video analysis
- learning materials
- spatial and temporal
- distance measure
- dimensionality reduction
- learning process
- spatio temporal
- combining information from multiple