RoMaV2: Harder, Better, Faster, Denser Feature Matching Model

tom_doerr · x · 2026-08-15

RoMaV2 is a new feature matching model designed for denser matching on image pairs. It supports arbitrary inputs, provides pixel coordinate conversion, and offers sampling capabilities. The project is open-sourced on GitHub with a link to the associated paper and usage examples.

Original post →

More from Research

Research channel →