04:00
2026-06-18
arxiv.org
computer-vision
RegimeVGGT: Layer-Wise Spatially Preserving Redundancy Removal for Visual Geometry Grounded Transformer
Researchers introduced RegimeVGGT, a training-free accelerator for Visual Geometry Grounded Transformer (VGGT) that achieves a 6.7x speedup over VGGT* at matched reconstruction quality by applying lay…