M^3: Dense Matching Meets Multi-View Foundation Models for Monocular Gaussian Splatting SLAM

2026-03-17

AI Summary

Monocular video-based pose estimation and geometric reconstruction in dynamic scenes often struggle to simultaneously achieve high accuracy and computational efficiency. To address this challenge, this work proposes a novel framework that integrates a multi-view foundation model with Gaussian Splatting SLAM. The approach introduces, for the first time, a dense matching head to generate pixel-level fine-grained correspondences, complemented by a dynamic region suppression mechanism and a cross-inference intrinsic parameter alignment strategy. These innovations significantly enhance tracking stability and geometric reconstruction quality. Evaluated on benchmarks such as ScanNet++, the method achieves state-of-the-art performance, reducing the ATE RMSE by 64.3% compared to VGGT-SLAM 2.0 and surpassing ARTDECO by 2.11 dB in PSNR.

Technology Category

• Computer Vision: Motion & Tracking
• Intelligent Robots: State Estimation
• Machine Learning: Multi-instance/Multi-view Learning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Streaming reconstruction from uncalibrated monocular video remains challenging, as it requires both high-precision pose estimation and computationally efficient online refinement in dynamic environments. While coupling 3D foundation models with SLAM frameworks is a promising paradigm, a critical bottleneck persists: most multi-view foundation models estimate poses in a feed-forward manner, yielding pixel-level correspondences that lack the requisite precision for rigorous geometric optimization. To address this, we present M^3, which augments the Multi-view foundation model with a dedicated Matching head to facilitate fine-grained dense correspondences and integrates it into a robust Monocular Gaussian Splatting SLAM. M^3 further enhances tracking stability by incorporating dynamic area suppression and cross-inference intrinsic alignment. Extensive experiments on diverse indoor and outdoor benchmarks demonstrate state-of-the-art accuracy in both pose estimation and scene reconstruction. Notably, M^3 reduces ATE RMSE by 64.3% compared to VGGT-SLAM 2.0 and outperforms ARTDECO by 2.11 dB in PSNR on the ScanNet++ dataset.

Problem

• monocular SLAM
• streaming reconstruction
• pose estimation
• dense matching
• multi-view foundation models

Innovation

• dense matching
• multi-view foundation models
• monocular SLAM
• Gaussian splatting
• pose estimation