BA-T: An Iterative Transformer for Two-View Bundle Adjustment

Arxiv CS.CV (computer vision)

The paper introduces BA-T, a feed-forward transformer model for iterative two-view bundle adjustment in 3D reconstruction. It utilizes deep cross-view attention to exchange information across images, avoiding heavy decoder stacks.

transformercomputer-vision3d-reconstruction

Arxiv CS.CV (computer vision)·Ganlin ZhangWeirong ChenDaniel CremersXi Wang

8 days ago