M3: Dense Matching Meets Multi-View Foundation Models for Monocular Gaussian Splatting SLAM
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. We present M3, a Monocular Gaussian Splatting SLAM that enhances Multi-view foundation models with a dedicated Matching head for fine-grained pose optimization. Our framework utilizes a unified inference strategy to jointly optimize historical and incoming frames, while integrating dynamic area suppression and cross-inference intrinsic alignment to maintain robustness. Extensive experiments on diverse indoor and outdoor benchmarks demonstrate state-of-the-art accuracy in both pose estimation and scene reconstruction. Notably, M3 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. Project page: https://city-super.github.io/M3.