D2-Net: A Trainable CNN for Joint Description and Detection of Local Features
Paper i proceeding, 2019

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.

3D reconstruction

deep learning

visual localization

machine learning

local features

Författare

Mihai Dusmanu

Ecole Normale Superieure (ENS)

INRIA Paris

ETH Zurich

Ignacio Rocco

Ecole Normale Superieure (ENS)

INRIA Paris

Tomas Pajdla

Ceske Vysoke Uceni Technicke v Praze

Marc Pollefeys

ETH Zurich

Microsoft

Josef Sivic

Ecole Normale Superieure (ENS)

Ceske Vysoke Uceni Technicke v Praze

INRIA Paris

Akihiko Torii

Tokyo Institute of Technology

Torsten Sattler

Chalmers, Elektroteknik, Signalbehandling och medicinsk teknik, Digitala bildsystem och bildanalys

IEEE / CVF Conference on Computer Vision and Pattern Recognition
Long Beach, USA,

Ämneskategorier

Robotteknik och automation

Datorseende och robotik (autonoma system)

Mer information

Skapat

2019-08-19