Global Structure-from-Motion Meets Feedforward Reconstruction

2026-05-25

AI Summary

This work addresses the limitations of traditional Structure-from-Motion (SfM) methods, which often fail in low-texture, minimally overlapping, or symmetric scenes, as well as the shortcomings of feedforward deep learning–based 3D reconstruction approaches in terms of scalability, accuracy, and robustness. To overcome these challenges, the paper proposes a novel framework that synergistically integrates classical optimization-based SfM with end-to-end scalable deep reconstruction models, enabling complementary strengths between the two paradigms. The proposed method achieves state-of-the-art performance across multiple benchmark datasets, significantly improving reconstruction success rate, accuracy, and robustness. The authors publicly release their code to facilitate further research in this direction.

Abstract

Structure-from-Motion -- the process of simultaneously estimating camera poses and 3D scene structure from a collection of images -- remains a central challenge in computer vision, with many open problems yet to be solved. Recent advances in feedforward 3D reconstruction have made significant strides in overcoming persistent failure cases of classical SfM methods, particularly in scenarios characterized by low texture, limited overlap, and symmetries. However, while feedforward approaches excel in these challenging conditions, they often face limitations regarding scalability, accuracy, or robustness, and typically fall short of classical methods in standard reconstruction settings. In this work, we systematically analyze these limitations and propose a new Structure-from-Motion pipeline by combining the respective strengths of classical and feedforward methods. Extensive experiments across multiple datasets show the benefits of our approach, achieving state-of-the-art results across a wide range of scenarios. We share our system as an open-source implementation at https://github.com/colmap/gluemap.

Problem

• Structure-from-Motion
• 3D reconstruction
• feedforward methods
• scalability
• robustness

Innovation

• Structure-from-Motion
• feedforward reconstruction
• 3D reconstruction
• hybrid pipeline
• scalability