SCOPE: Scale-Consistent One-Pass Estimation of 3D Geometry

2026-06-19

AI Summary

Existing methods struggle to simultaneously achieve geometric accuracy and temporal consistency in long-duration monocular video reconstruction. This work proposes an end-to-end neural network that generates affine-invariant 3D point maps through a single forward pass, enabling scale-consistent reconstruction within a unified reference frame. The approach introduces three core innovations: viewpoint-invariant geometric alignment, appearance-invariant learning, and frequency-modulated localization, which collectively facilitate consistent modeling across exponentially varying time scales and support robust extrapolation over ultra-long sequences. Experiments on datasets such as ScanNet demonstrate a 24.2% reduction in point map error and a 34.9% decrease in temporal alignment error, significantly enhancing reconstruction robustness under complex camera trajectories and varying illumination conditions.

Abstract

We present SCOPE (Scale-Consistent One-Pass Estimation of 3D Geometry), a novel approach for estimating 3D geometry from extended monocular video sequences, where existing methods struggle to maintain both geometric accuracy and temporal consistency across hundreds of frames. Our approach generates affine-invariant 3D point maps with shared parameters across entire sequences, enabling consistent scale-invariant representations. We introduce three key innovations: viewpoint-invariant geometry aligning multi-perspective points in a unified reference frame; appearance-invariant learning enforcing consistency across exponential timescales; and frequency-modulated positioning enabling extrapolation to sequences vastly exceeding training length. Experiments across diverse datasets demonstrate significant improvements, reducing relative point map error by 24.2% and temporal alignment error by 34.9% on ScanNet compared to state-of-the-art methods. Our approach handles challenging scenarios with complex camera trajectories and lighting variations while efficiently processing extended sequences in a single pass. Project page: https://scope3d.github.io/.

Problem

• 3D geometry estimation
• monocular video
• temporal consistency
• geometric accuracy
• scale consistency

Innovation

• scale-consistent
• one-pass estimation
• viewpoint-invariant geometry
• appearance-invariant learning
• frequency-modulated positioning