Enriching Phrases with Coupled Pixel and Object Contexts for Panoptic Narrative Grounding.
Tianrui HuiZihan DingJunshi HuangXiaoming WeiXiaolin WeiJiao DaiJizhong HanSi LiuPublished in: CoRR (2023)
Keyphrases
- d objects
- object model
- graph cut segmentation
- complex objects
- moving objects
- viewpoint
- reference object
- input image
- object segmentation
- background pixels
- object level
- pixel level
- search engine
- data objects
- image features
- natural language
- spatial relationships
- image frames
- intensity values
- object tracking
- detected edges
- gradient orientation