RelTransformer: Balancing the Visual Relationship Detection from Local Context, Scene and Memory.
Jun ChenAniket AgarwalSherif AbdelkarimDeyao ZhuMohamed ElhoseinyPublished in: CoRR (2021)
Keyphrases
- visual scene
- visual data
- detection rate
- video sequences
- visual context
- contextual information
- input image
- detection algorithm
- detection method
- visual appearance
- observed scene
- visual environment
- d scene
- visual features
- object detection
- context aware
- spatial relations
- complex scenes
- video scene
- low level
- moving objects
- thermal images
- detecting and tracking multiple
- scene categorization
- crowded scenes
- computer vision
- text localization and recognition
- visual analysis
- outdoor scenes
- scene analysis
- real scenes
- multiple images
- main memory
- visual information
- false positives
- single image
- high level