Dense 3D Reconstruction
3D Reconstruction from traditional SfM, MVS is time consuming and involves complex intermediary steps. VGGT (Visual Geometry Grounded Transformer) outperforms DUSt3R and MASt3R in multiple benchmarks achieving SOTA results.
MASt3R (Multi View Stereo 3D Reconstruction) is a 3D aware image matches that grounds image matching as a 3D task to establish better correspondence. In this article, we will understand
