RoMa: Robust Dense Feature Matching
By Johan Edstedt, Qiyu Sun, Georg Bökman, Mårten Wadenbäck, Michael Felsberg. Computer Vision and Pattern Recognition, 2023
381 citations recorded by the source corpus.
Feature matching is an important computer vision task that involves estimating correspondences between two images of a 3D scene, and dense methods estimate all such correspondences. The aim is to learn a robust model, i.e., a model able to match under challenging real-world changes. In this work, we propose such a model, leveraging frozen pretrained features from the foundation model DINOv2. Al-though these features are significantly more robust than local features trained from scratch, they are inherently coarse. We therefore combine them with specialized ConvNet fine features, creating a precisely localizable feature pyramid. To further improve robustness, we propose a tailored transformer match decoder that predicts anchor probabilities, which enables it to express multimodality. Finally, we propose an improved loss formulation through regression-by-classification with subsequent robust regression. We conduct a comprehensive set of experiments that show that our method, RoMa, achieves significant gains, setting a new state-of-the-art. In particular, we achieve a 36% improvement on the extremely challenging WxBS benchmark. Code is provided at github.com/Parskatt/RoMa.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition
View Paper: https://www.semanticscholar.org/paper/4217d99affbeab7f9fff8d13e3cf15b40e1f794e
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Christian Rupprecht
- David Novotný
- Jérôme Revaud
- Andrea Vedaldi
- Vincent Leroy
- Yohann Cabon
- Andrew Zisserman
- Lior Wolf
In Computer Vision and Pattern Recognition
- Andrea Vedaldi
- Luc Van Gool
- Christian Rupprecht
- Andrew Zisserman
- Jitendra Malik
- Ming-Hsuan Yang
- David Novotný
- Dacheng Tao
Citations in the corpus
- DUSt3R: Geometric 3D Vision Made Easy (2023, methodology): “These improvements encompass advanced feature description [26, 28, 97, 135, 167], more accurate image matching [3, 17, 60, 82, 100, 120, 126, 145], featuremetric refinement [59], and neural bundle adjustment [58, 153].”
- VGGT: Visual Geometry Grounded Transformer (2025, positive): “Measured in AUC (higher is better). evaluation hyperparameters ( e.g ., the number of matches, RANSAC thresholds) from Roma [21].”
- Grounding Image Matching in 3D with MASt3R (2024, result): “In contrast to keypoint-based approaches, semi-dense [11,16,43,46,82,85] and dense approaches [27,28,29,58,92,93,94,122] offer a different paradigm for establishing image correspondences, considering all possible pixel associations.”