D2-Net: A Trainable CNN for Joint Description and Detection of Local Features

Mihai Dusmanu, Ignacio Rocco, Tomas Pajdla, Marc Pollefeys, Josef Sivic, Akihiko Torii, Torsten Sattler

In this work we address the problem of finding reliable pixel-level correspondences under difficult imaging conditions. We propose an approach where a single convolutional neural network plays a dual role: It is simultaneously a dense feature descriptor and a feature detector. By postponing the detection to a later stage, the obtained keypoints are more stable than their traditional counterparts based on early detection of low-level structures. We show that this model can be trained using pixel correspondences extracted from readily available large-scale SfM reconstructions, without any further annotations. The proposed method obtains state-of-the-art performance on both the difficult Aachen Day-Night localization dataset and the InLoc indoor localization benchmark, as well as competitive performance on other benchmarks for image matching and 3D reconstruction.

• Microsoft
• Microsoft
• Massachusetts Institute of Technology
• National Bureau of Economic Research (NBER)
• University of Pennsylvania
• Microsoft
• Microsoft
• Carnegie Mellon University
• Stanford University
• University of California, Berkeley
• Microsoft
• ETH Zurich
• University of Zurich
• Microsoft