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. Figure 1: Examples of matches obtained by the D2-Net method. The proposed method can find image correspondences even under significant appearance differences caused by strong changes in illumination such as day-to-night, changes in depiction style or under image degradation caused by motion blur.ciently via (approximate) nearest neighbor search [37] and the Euclidean distance. Sparse features offer a memory efficient representation and thus enable approaches such as Structure-from-Motion (SfM) [21,53] or visual localization [26,47,58] to scale. The keypoint detector typically considers low-level image information such as corners [19] or blob-like structures [30,32]. As such, local features can

No abstract