AMB3R: Accurate Feed-forward Metric-scale 3D Reconstruction with Backend
This addresses the need for accurate and efficient 3D reconstruction in computer vision applications, offering a versatile model that extends to tasks like visual odometry without fine-tuning.
The paper tackles the problem of dense 3D reconstruction at metric scale by introducing AMB3R, a multi-view feed-forward model that uses a sparse volumetric backend for geometric reasoning. It achieves state-of-the-art performance in camera pose, depth, and metric-scale estimation, surpassing optimization-based SLAM and SfM methods on benchmarks.
We present AMB3R, a multi-view feed-forward model for dense 3D reconstruction on a metric-scale that addresses diverse 3D vision tasks. The key idea is to leverage a sparse, yet compact, volumetric scene representation as our backend, enabling geometric reasoning with spatial compactness. Although trained solely for multi-view reconstruction, we demonstrate that AMB3R can be seamlessly extended to uncalibrated visual odometry (online) or large-scale structure from motion without the need for task-specific fine-tuning or test-time optimization. Compared to prior pointmap-based models, our approach achieves state-of-the-art performance in camera pose, depth, and metric-scale estimation, 3D reconstruction, and even surpasses optimization-based SLAM and SfM methods with dense reconstruction priors on common benchmarks.