2 datasets found

The dataset used in the paper is MegaDepth and NYU datasets for training and testing the proposed method.

Feature matching is a fundamental problem for many computer vision tasks, such as object recognition, structure from motion, and simultaneous localization and mapping.