RoMa v2: Harder Better Faster Denser Feature Matching

TL;DR

RoMa v2 introduces a series of architectural and training improvements for dense feature matching, achieving state-of-the-art accuracy and robustness in complex real-world scenarios while enhancing speed and memory efficiency.

Contribution

The paper presents a novel matching architecture, a new loss function, a two-stage pipeline, and integration with DINOv3 to significantly improve dense feature matching performance.

Findings

Sets new state-of-the-art accuracy in dense matching

Improves robustness to complex real-world scenarios

Reduces training and refinement memory usage

Abstract

Dense feature matching aims to estimate all correspondences between two images of a 3D scene and has recently been established as the gold-standard due to its high accuracy and robustness. However, existing dense matchers still fail or perform poorly for many hard real-world scenarios, and high-precision models are often slow, limiting their applicability. In this paper, we attack these weaknesses on a wide front through a series of systematic improvements that together yield a significantly better model. In particular, we construct a novel matching architecture and loss, which, combined with a curated diverse training distribution, enables our model to solve many complex matching tasks. We further make training faster through a decoupled two-stage matching-then-refinement pipeline, and at the same time, significantly reduce refinement memory usage through a custom CUDA kernel. Finally, we leverage the recent DINOv3 foundation model along with multiple other insights to make the model more robust and unbiased. In our extensive set of experiments we show that the resulting novel matcher sets a new state-of-the-art, being significantly more accurate than its predecessors. Code is available at https://github.com/Parskatt/romav2