Search papers, labs, and topics across Lattice.
The paper introduces Edge-based Pose Optimization (EPO), a novel geometric optimization framework that enhances the accuracy of Structure-from-Motion reconstructions produced by 3D Foundation Models without relying on traditional feature extraction and matching. By utilizing edge map alignment as a proxy for geometric optimization, EPO avoids the computational overhead associated with track construction, thereby preserving the rapid inference capabilities of these models. Extensive evaluations reveal that EPO not only matches but often surpasses the performance of conventional Bundle Adjustment methods while significantly reducing runtime and memory requirements, making it accessible for consumer-grade hardware.
EPO achieves high geometric accuracy in 3D reconstructions without the computational burden of traditional feature extraction, making advanced 3D modeling feasible on consumer-grade devices.
We introduce \textbf{Edge-based Pose Optimization (EPO)}, a trackless geometric optimization framework specifically designed to boost the Structure-from-Motion reconstructions generated by 3D Foundation Models. These models achieve rapid inference by bypassing the time-consuming feature extraction and matching stages of traditional pipelines, where explicit correspondences between each 3D point and multiple images, referred to as tracks, are established. However, their geometric accuracy currently falls short of traditional pipelines. While this can be addressed in a post-processing step via Bundle Adjustment-like refinement, doing so requires extracting feature tracks, thus defeating the original speed advantage. Instead, our fully differentiable framework uses edge map alignment as a proxy for geometric optimization, avoiding feature extraction and track construction entirely. Through extensive evaluation across multiple datasets and tasks, we demonstrate that EPO matches or outperforms Bundle Adjustment-like methods while requiring significantly lower runtime and memory. Notably, its reduced memory footprint makes EPO suitable for consumer-grade hardware, where competing refinement methods cannot run.
