Abstract: Visual grounding (VG) is a critical task that seeks to identify and localize a specific visual region within a given image based on a corresponding referring expression. Existing approaches ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results