Automatic Annotation of Airborne Images by Label Propagation Based on a Bayesian-CRF Model

Zhuo Xiangyu, Friedrich Fraundorfer, Franz Kurz, Peter Reinartz

Publikation: Beitrag in einer FachzeitschriftArtikelForschungBegutachtung

Abstract

The tremendous advances in deep neural networks have demonstrated the superiority of deep learning techniques for applications such as object recognition or image classification. Nevertheless, deep learning-based methods usually require a large amount of training data, which mainly comes from manual annotation and is quite labor-intensive. In order to reduce the amount of manual work required for generating enough training data, we hereby propose to leverage existing labeled data to generate image annotations automatically. Specifically, the pixel labels are firstly transferred from one image modality to another image modality via geometric transformation to create initial image annotations, and then additional information (eg, height measurements) is incorporated for Bayesian inference to update the labeling beliefs. Finally, the updated label assignments are optimized with a fully connected conditional random field (CRF), yielding refined labeling for all pixels in the image. The proposed approach is tested on two different scenarios, ie,(1) label propagation from annotated aerial imagery to unmanned aerial vehicle (UAV) imagery and (2) label propagation from map database to aerial imagery. In each scenario, the refined image labels are used as pseudo-ground truth data for training a convolutional neural network (CNN). Results demonstrate that our model is able to produce accurate label assignments even around complex object boundaries; besides, the generated image labels can be effectively leveraged for training CNNs and achieve comparable classification accuracy as manual image annotations, more specifically, the per …
Originalspracheenglisch
Aufsatznummer145
FachzeitschriftRemote Sensing
Jahrgang11
Ausgabenummer2
DOIs
PublikationsstatusVeröffentlicht - 2019

Fingerprint

imagery
pixel
learning
image classification
labor
labelling
method
vehicle

Dies zitieren

Automatic Annotation of Airborne Images by Label Propagation Based on a Bayesian-CRF Model. / Xiangyu, Zhuo; Fraundorfer, Friedrich; Kurz, Franz; Reinartz, Peter.

in: Remote Sensing , Jahrgang 11, Nr. 2, 145, 2019.

Publikation: Beitrag in einer FachzeitschriftArtikelForschungBegutachtung

@article{4f85be5e56aa4b1aaeb35318601871cb,
title = "Automatic Annotation of Airborne Images by Label Propagation Based on a Bayesian-CRF Model",
abstract = "The tremendous advances in deep neural networks have demonstrated the superiority of deep learning techniques for applications such as object recognition or image classification. Nevertheless, deep learning-based methods usually require a large amount of training data, which mainly comes from manual annotation and is quite labor-intensive. In order to reduce the amount of manual work required for generating enough training data, we hereby propose to leverage existing labeled data to generate image annotations automatically. Specifically, the pixel labels are firstly transferred from one image modality to another image modality via geometric transformation to create initial image annotations, and then additional information (eg, height measurements) is incorporated for Bayesian inference to update the labeling beliefs. Finally, the updated label assignments are optimized with a fully connected conditional random field (CRF), yielding refined labeling for all pixels in the image. The proposed approach is tested on two different scenarios, ie,(1) label propagation from annotated aerial imagery to unmanned aerial vehicle (UAV) imagery and (2) label propagation from map database to aerial imagery. In each scenario, the refined image labels are used as pseudo-ground truth data for training a convolutional neural network (CNN). Results demonstrate that our model is able to produce accurate label assignments even around complex object boundaries; besides, the generated image labels can be effectively leveraged for training CNNs and achieve comparable classification accuracy as manual image annotations, more specifically, the per …",
author = "Zhuo Xiangyu and Friedrich Fraundorfer and Franz Kurz and Peter Reinartz",
year = "2019",
doi = "10.3390/rs11020145",
language = "English",
volume = "11",
journal = "Remote Sensing",
issn = "2072-4292",
publisher = "MDPI AG",
number = "2",

}

TY - JOUR

T1 - Automatic Annotation of Airborne Images by Label Propagation Based on a Bayesian-CRF Model

AU - Xiangyu, Zhuo

AU - Fraundorfer, Friedrich

AU - Kurz, Franz

AU - Reinartz, Peter

PY - 2019

Y1 - 2019

N2 - The tremendous advances in deep neural networks have demonstrated the superiority of deep learning techniques for applications such as object recognition or image classification. Nevertheless, deep learning-based methods usually require a large amount of training data, which mainly comes from manual annotation and is quite labor-intensive. In order to reduce the amount of manual work required for generating enough training data, we hereby propose to leverage existing labeled data to generate image annotations automatically. Specifically, the pixel labels are firstly transferred from one image modality to another image modality via geometric transformation to create initial image annotations, and then additional information (eg, height measurements) is incorporated for Bayesian inference to update the labeling beliefs. Finally, the updated label assignments are optimized with a fully connected conditional random field (CRF), yielding refined labeling for all pixels in the image. The proposed approach is tested on two different scenarios, ie,(1) label propagation from annotated aerial imagery to unmanned aerial vehicle (UAV) imagery and (2) label propagation from map database to aerial imagery. In each scenario, the refined image labels are used as pseudo-ground truth data for training a convolutional neural network (CNN). Results demonstrate that our model is able to produce accurate label assignments even around complex object boundaries; besides, the generated image labels can be effectively leveraged for training CNNs and achieve comparable classification accuracy as manual image annotations, more specifically, the per …

AB - The tremendous advances in deep neural networks have demonstrated the superiority of deep learning techniques for applications such as object recognition or image classification. Nevertheless, deep learning-based methods usually require a large amount of training data, which mainly comes from manual annotation and is quite labor-intensive. In order to reduce the amount of manual work required for generating enough training data, we hereby propose to leverage existing labeled data to generate image annotations automatically. Specifically, the pixel labels are firstly transferred from one image modality to another image modality via geometric transformation to create initial image annotations, and then additional information (eg, height measurements) is incorporated for Bayesian inference to update the labeling beliefs. Finally, the updated label assignments are optimized with a fully connected conditional random field (CRF), yielding refined labeling for all pixels in the image. The proposed approach is tested on two different scenarios, ie,(1) label propagation from annotated aerial imagery to unmanned aerial vehicle (UAV) imagery and (2) label propagation from map database to aerial imagery. In each scenario, the refined image labels are used as pseudo-ground truth data for training a convolutional neural network (CNN). Results demonstrate that our model is able to produce accurate label assignments even around complex object boundaries; besides, the generated image labels can be effectively leveraged for training CNNs and achieve comparable classification accuracy as manual image annotations, more specifically, the per …

U2 - 10.3390/rs11020145

DO - 10.3390/rs11020145

M3 - Article

VL - 11

JO - Remote Sensing

JF - Remote Sensing

SN - 2072-4292

IS - 2

M1 - 145

ER -