SEAR: Simple and Efficient Adaptation of Visual Geometric Transformers for RGB+Thermal 3D Reconstruction

SEAR: Simple and Efficient Adaptation of Visual Geometric Transformers for RGB+Thermal 3D Reconstruction explores A simple fine-tuning strategy that adapts existing visual geometry models for accurate RGB-thermal 3D reconstruction, outperforming state-of-the-art even in challenging conditions.. Commercial viability score: 8/10 in 3D Reconstruction.