DAGE: Dual-Stream Architecture for Efficient and Fine-Grained Geometry Estimation

TL;DR

DAGE introduces a dual-stream transformer architecture that efficiently estimates high-resolution, view-consistent geometry and camera poses from uncalibrated multi-view videos, balancing global coherence and fine detail.

Contribution

The paper presents DAGE, a novel dual-stream transformer that disentangles global and local features for scalable, accurate geometry and pose estimation from high-resolution videos.

Findings

Achieves state-of-the-art results in video geometry estimation.

Supports inputs up to 2K resolution with practical inference costs.

Produces sharp depth maps and maintains cross-view consistency.

Abstract

Estimating accurate, view-consistent geometry and camera poses from uncalibrated multi-view/video inputs remains challenging - especially at high spatial resolutions and over long sequences. We present DAGE, a dual-stream transformer whose main novelty is to disentangle global coherence from fine detail. A low-resolution stream operates on aggressively downsampled frames with alternating frame/global attention to build a view-consistent representation and estimate cameras efficiently, while a high-resolution stream processes the original images per-frame to preserve sharp boundaries and small structures. A lightweight adapter fuses these streams via cross-attention, injecting global context without disturbing the pretrained single-frame pathway. This design scales resolution and clip length independently, supports inputs up to 2K, and maintains practical inference cost. DAGE delivers sharp depth/pointmaps, strong cross-view consistency, and accurate poses, establishing new state-of-the-art results for video geometry estimation and multi-view reconstruction.