DFormerv2: Geometry Self-Attention for RGBD Semantic Segmentation Paper • 2504.04701 • Published Apr 7 • 2