Remotely Operated Vehicles (ROVs) are pivotal for subsea inspection, maintenance and repair (IMR) operations. Increased autonomy in such IMR operations may constitute significant improvements in HSE (Health- Safety, Environment) and cost-effectiveness of operations. Localization is an enabling technology for autonomy. In this paper, we extend a localization framework called Vind (developed by SINTEF) from 2D mobile robot localization to 3D ROV localization. Vind framework keywords include modularity, scalability and reconfigurability. This includes that the framework facilitates easy change of, e.g., system modules, integration filters and sensors. We experimentally validate the system with real sensor data from an ROV mission. The localization systems performs even after we add additional drop-outs and varying noise characteristics to the sensor data. In the paper, we also discuss aspects concerning autonomy in IMR operations with unmanned underwater vehicles.