Concept视觉接地 / Visual Grounding
Visual Grounding
The ability of vision-language models to map natural language referring expressions to pixel regions in images. It is the foundation for referential understanding, object detection and spatial reasoning.
Timeline (last 90 days)
Oct 19
Visual grounding refers to the process by which a model maps referring expressions in natural language to specific pixel regions in an image.
Unverified50%