IMAGE-TO-IMAGE TRANSLATION for ENHANCED FEATURE MATCHING, IMAGE RETRIEVAL and VISUAL LOCALIZATION
Paper i proceeding, 2019

The performance of machine learning and deep learning algorithms for image analysis depends significantly on the quantity and quality of the training data. The generation of annotated training data is often costly, time-consuming and laborious. Data augmentation is a powerful option to overcome these drawbacks. Therefore, we augment training data by rendering images with arbitrary poses from 3D models to increase the quantity of training images. These training images usually show artifacts and are of limited use for advanced image analysis. Therefore, we propose to use image-to-image translation to transform images from a rendered domain to a captured domain. We show that translated images in the captured domain are of higher quality than the rendered images. Moreover, we demonstrate that image-to-image translation based on rendered 3D models enhances the performance of common computer vision tasks, namely feature matching, image retrieval and visual localization. The experimental results clearly show the enhancement on translated images over rendered images for all investigated tasks. In addition to this, we present the advantages utilizing translated images over exclusively captured images for visual localization.

Convolutional Neural Networks

Generative Adversarial Networks

3D Models

Image Retrieval

Image-to-Image Translation

Feature Matching

Data Augmentation

Visual Localization

Författare

M. S. Mueller

Karlsruher Institut für Technologie (KIT)

Torsten Sattler

Chalmers, Elektroteknik, Signalbehandling och medicinsk teknik

Marc Pollefeys

Eidgenössische Technische Hochschule Zürich (ETH)

Microsoft

B. Jutzi

Karlsruher Institut für Technologie (KIT)

ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences

21949042 (ISSN) 21949050 (eISSN)

Vol. 4 2/W7 111-119

1st Photogrammetric Image Analysis and Munich Remote Sensing Symposium, PIA 2019+MRSS 2019
Munich, Germany,

Ämneskategorier

Mediateknik

Datorseende och robotik (autonoma system)

Medicinsk bildbehandling

DOI

10.5194/isprs-annals-IV-2-W7-111-2019

Mer information

Senast uppdaterat

2020-10-16