Geometry-Grounded Video Depth Estimation

Our goal is to infer temporally and geometrically consistent depth maps that lift monocular video clips into 3D. To this end, we develop $\text{G2}$, a geometry-grounded feed-forward framework that fills in three missing pieces in video depth estimation, $\text{i}.\text{e}.$, cross-frame modelling, geometric supervision, and data scaling. First, instead of most existing efforts relying on fixed-location feature aggregation or externally estimated correspondences, we propose $\text{Motion-Adaptive Implicit Alignment}$ to adaptively locate a sparse set of query-relevant features for efficient cross-frame information aggregation under camera and object motion. Second, we introduce $\text{Explicit 3D Supervision}$ to directly constrain intra-frame structure and inter-frame geometry after back-projection of depth maps. Third, to scale geometric supervision to unlabeled videos, we develop $\text{Multi-View Geometric Distillation}$, which distills geometric knowledge from a reconstruction teacher to provide pseudo 3D supervision, yielding a more scalable solution. Extensive experiments on standard benchmarks demonstrate that G2 achieves state-of-the-art performance on both zero-shot video depth estimation and geometric reconstruction. We will release the code and model.