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

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21730%2F19%3A00337394" target="_blank" >RIV/68407700:21730/19:00337394 - isvavai.cz</a>

Výsledek na webu

<a href="https://doi.org/10.1109/CVPR.2019.00828" target="_blank" >https://doi.org/10.1109/CVPR.2019.00828</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/CVPR.2019.00828" target="_blank" >10.1109/CVPR.2019.00828</a>

Jazyk výsledku

angličtina

Název v původním jazyce

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

Popis výsledku v původním jazyce

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.

Název v anglickém jazyce

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

Popis výsledku anglicky

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.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

Výsledek vznikl pri realizaci vícero projektů. Více informací v záložce Projekty.

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2019

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

CVPR 2019: Proceedings of the 2019 IEEE Conference on Computer Vision and Pattern Recognition

ISBN

978-1-7281-3294-5

ISSN

1063-6919

e-ISSN

2575-7075

Počet stran výsledku

10

Strana od-do

8084-8093

Název nakladatele

IEEE

Místo vydání

—

Místo konání akce

Long Beach

Datum konání akce

15. 6. 2019

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

000542649301070