News
The data generators with the first approach, augmenting Visual Genome’s scene graphs with depth and segmentation annotation from Depth Anything V2 and SAM-2, helped them create 1.5 million ...
Results that may be inaccessible to you are currently showing.
Hide inaccessible results