You are here: Home → Proceedings
Document Actions

Citation and metadata

Back to the article.

Recommended citation

Knitt M, Schyga J, Adamanov A, Hinckeldeyn J, Kreutzfeldt J (2022). Estimating the Pose of a Euro Pallet with an RGB Camera based on Synthetic Training Data. Logistics Journal : Proceedings, Vol. 2022. (urn:nbn:de:0009-14-55900)

Download Citation

Endnote

%0 Journal Article
%T Estimating the Pose of a Euro Pallet with an RGB Camera based on Synthetic Training Data
%A Knitt, Markus
%A Schyga, Jakob
%A Adamanov, Asan
%A Hinckeldeyn, Johannes
%A Kreutzfeldt, Jochen
%J Logistics Journal : Proceedings
%D 2022
%V 2022
%N 18
%@ 2192-9084
%F knitt2022
%X Estimating the pose of a pallet and other logistics objects is crucial for various use cases, such as automatized material handling or tracking. Innovations in computer vision, computing power, and machine learning open up new opportunities for device-free localization based on cameras and neural networks. Large image datasets with annotated poses are required for training the network. Manual annotation, especially of 6D poses, is an extremely labor-intensive process. Hence, newer approaches often leverage synthetic training data to automatize the process of generating annotated image datasets. In this work, the generation of synthetic training data for 6D pose estimation of pallets is presented. The data is then used to train the Deep Object Pose Estimation (DOPE) algorithm [1]. The experimental validation of the algorithm proves that the 6D pose estimation of a standardized Euro pallet with a Red-Green-Blue (RGB) camera is feasible. The comparison of the results from three varying datasets under different lighting conditions shows the relevance of an appropriate dataset design to achieve an accurate and robust localization. The quantitative evaluation shows an average position error of less than 20 cm for the preferred dataset. The validated training dataset and a photorealistic model of a Euro pallet are publicly provided [2].
%L 620
%K 6D pose estimation
%K 6D-Posenschätzung
%K DOPE algorithm
%K DOPE-Algorithmus
%K Europalette
%K RGB camera
%K RGB-Kamera
%K Synthetischer Trainingsdatensatz
%K euro pallet
%K synthetic training dataset
%R 10.2195/lj_proc_knitt_en_202211_01
%U http://nbn-resolving.de/urn:nbn:de:0009-14-55900
%U http://dx.doi.org/10.2195/lj_proc_knitt_en_202211_01
Download

Bibtex

@Article{knitt2022,
  author = 	"Knitt, Markus
		and Schyga, Jakob
		and Adamanov, Asan
		and Hinckeldeyn, Johannes
		and Kreutzfeldt, Jochen",
  title = 	"Estimating the Pose of a Euro Pallet with an RGB Camera based on Synthetic Training Data",
  journal = 	"Logistics Journal : Proceedings",
  year = 	"2022",
  volume = 	"2022",
  number = 	"18",
  keywords = 	"6D pose estimation; 6D-Posensch{\"a}tzung; DOPE algorithm; DOPE-Algorithmus; Europalette; RGB camera; RGB-Kamera; Synthetischer Trainingsdatensatz; euro pallet; synthetic training dataset",
  abstract = 	"Estimating the pose of a pallet and other logistics objects is crucial for various use cases, such as automatized material handling or tracking. Innovations in computer vision, computing power, and machine learning open up new opportunities for device-free localization based on cameras and neural networks. Large image datasets with annotated poses are required for training the network. Manual annotation, especially of 6D poses, is an extremely labor-intensive process. Hence, newer approaches often leverage synthetic training data to automatize the process of generating annotated image datasets. In this work, the generation of synthetic training data for 6D pose estimation of pallets is presented. The data is then used to train the Deep Object Pose Estimation (DOPE) algorithm [1]. The experimental validation of the algorithm proves that the 6D pose estimation of a standardized Euro pallet with a Red-Green-Blue (RGB) camera is feasible. The comparison of the results from three varying datasets under different lighting conditions shows the relevance of an appropriate dataset design to achieve an accurate and robust localization. The quantitative evaluation shows an average position error of less than 20 cm for the preferred dataset. The validated training dataset and a photorealistic model of a Euro pallet are publicly provided [2].",
  issn = 	"2192-9084",
  doi = 	"10.2195/lj_proc_knitt_en_202211_01",
  url = 	"http://nbn-resolving.de/urn:nbn:de:0009-14-55900"
}
Download

RIS

TY  - JOUR
AU  - Knitt, Markus
AU  - Schyga, Jakob
AU  - Adamanov, Asan
AU  - Hinckeldeyn, Johannes
AU  - Kreutzfeldt, Jochen
PY  - 2022
DA  - 2022//
TI  - Estimating the Pose of a Euro Pallet with an RGB Camera based on Synthetic Training Data
JO  - Logistics Journal : Proceedings
VL  - 2022
IS  - 18
KW  - 6D pose estimation
KW  - 6D-Posenschätzung
KW  - DOPE algorithm
KW  - DOPE-Algorithmus
KW  - Europalette
KW  - RGB camera
KW  - RGB-Kamera
KW  - Synthetischer Trainingsdatensatz
KW  - euro pallet
KW  - synthetic training dataset
AB  - Estimating the pose of a pallet and other logistics objects is crucial for various use cases, such as automatized material handling or tracking. Innovations in computer vision, computing power, and machine learning open up new opportunities for device-free localization based on cameras and neural networks. Large image datasets with annotated poses are required for training the network. Manual annotation, especially of 6D poses, is an extremely labor-intensive process. Hence, newer approaches often leverage synthetic training data to automatize the process of generating annotated image datasets. In this work, the generation of synthetic training data for 6D pose estimation of pallets is presented. The data is then used to train the Deep Object Pose Estimation (DOPE) algorithm [1]. The experimental validation of the algorithm proves that the 6D pose estimation of a standardized Euro pallet with a Red-Green-Blue (RGB) camera is feasible. The comparison of the results from three varying datasets under different lighting conditions shows the relevance of an appropriate dataset design to achieve an accurate and robust localization. The quantitative evaluation shows an average position error of less than 20 cm for the preferred dataset. The validated training dataset and a photorealistic model of a Euro pallet are publicly provided [2].
SN  - 2192-9084
UR  - http://nbn-resolving.de/urn:nbn:de:0009-14-55900
DO  - 10.2195/lj_proc_knitt_en_202211_01
ID  - knitt2022
ER  -
Download

Wordbib

<?xml version="1.0" encoding="UTF-8"?>
<b:Sources SelectedStyle="" xmlns:b="http://schemas.openxmlformats.org/officeDocument/2006/bibliography"  xmlns="http://schemas.openxmlformats.org/officeDocument/2006/bibliography" >
<b:Source>
<b:Tag>knitt2022</b:Tag>
<b:SourceType>ArticleInAPeriodical</b:SourceType>
<b:Year>2022</b:Year>
<b:PeriodicalTitle>Logistics Journal : Proceedings</b:PeriodicalTitle>
<b:Volume>2022</b:Volume>
<b:Issue>18</b:Issue>
<b:Url>http://nbn-resolving.de/urn:nbn:de:0009-14-55900</b:Url>
<b:Url>http://dx.doi.org/10.2195/lj_proc_knitt_en_202211_01</b:Url>
<b:Author>
<b:Author><b:NameList>
<b:Person><b:Last>Knitt</b:Last><b:First>Markus</b:First></b:Person>
<b:Person><b:Last>Schyga</b:Last><b:First>Jakob</b:First></b:Person>
<b:Person><b:Last>Adamanov</b:Last><b:First>Asan</b:First></b:Person>
<b:Person><b:Last>Hinckeldeyn</b:Last><b:First>Johannes</b:First></b:Person>
<b:Person><b:Last>Kreutzfeldt</b:Last><b:First>Jochen</b:First></b:Person>
</b:NameList></b:Author>
</b:Author>
<b:Title>Estimating the Pose of a Euro Pallet with an RGB Camera based on Synthetic Training Data</b:Title>
<b:Comments>Estimating the pose of a pallet and other logistics objects is crucial for various use cases, such as automatized material handling or tracking. Innovations in computer vision, computing power, and machine learning open up new opportunities for device-free localization based on cameras and neural networks. Large image datasets with annotated poses are required for training the network. Manual annotation, especially of 6D poses, is an extremely labor-intensive process. Hence, newer approaches often leverage synthetic training data to automatize the process of generating annotated image datasets. In this work, the generation of synthetic training data for 6D pose estimation of pallets is presented. The data is then used to train the Deep Object Pose Estimation (DOPE) algorithm [1]. The experimental validation of the algorithm proves that the 6D pose estimation of a standardized Euro pallet with a Red-Green-Blue (RGB) camera is feasible. The comparison of the results from three varying datasets under different lighting conditions shows the relevance of an appropriate dataset design to achieve an accurate and robust localization. The quantitative evaluation shows an average position error of less than 20 cm for the preferred dataset. The validated training dataset and a photorealistic model of a Euro pallet are publicly provided [2].</b:Comments>
</b:Source>
</b:Sources>
Download

ISI

PT Journal
AU Knitt, M
   Schyga, J
   Adamanov, A
   Hinckeldeyn, J
   Kreutzfeldt, J
TI Estimating the Pose of a Euro Pallet with an RGB Camera based on Synthetic Training Data
SO Logistics Journal : Proceedings
PY 2022
VL 2022
IS 18
DI 10.2195/lj_proc_knitt_en_202211_01
DE 6D pose estimation; 6D-Posenschätzung; DOPE algorithm; DOPE-Algorithmus; Europalette; RGB camera; RGB-Kamera; Synthetischer Trainingsdatensatz; euro pallet; synthetic training dataset
AB Estimating the pose of a pallet and other logistics objects is crucial for various use cases, such as automatized material handling or tracking. Innovations in computer vision, computing power, and machine learning open up new opportunities for device-free localization based on cameras and neural networks. Large image datasets with annotated poses are required for training the network. Manual annotation, especially of 6D poses, is an extremely labor-intensive process. Hence, newer approaches often leverage synthetic training data to automatize the process of generating annotated image datasets. In this work, the generation of synthetic training data for 6D pose estimation of pallets is presented. The data is then used to train the Deep Object Pose Estimation (DOPE) algorithm [1]. The experimental validation of the algorithm proves that the 6D pose estimation of a standardized Euro pallet with a Red-Green-Blue (RGB) camera is feasible. The comparison of the results from three varying datasets under different lighting conditions shows the relevance of an appropriate dataset design to achieve an accurate and robust localization. The quantitative evaluation shows an average position error of less than 20 cm for the preferred dataset. The validated training dataset and a photorealistic model of a Euro pallet are publicly provided [2].
ER
Download

Mods

<mods>
  <titleInfo>
    <title>Estimating the Pose of a Euro Pallet with an RGB Camera based on Synthetic Training Data</title>
  </titleInfo>
  <name type="personal">
    <namePart type="family">Knitt</namePart>
    <namePart type="given">Markus</namePart>
  </name>
  <name type="personal">
    <namePart type="family">Schyga</namePart>
    <namePart type="given">Jakob</namePart>
  </name>
  <name type="personal">
    <namePart type="family">Adamanov</namePart>
    <namePart type="given">Asan</namePart>
  </name>
  <name type="personal">
    <namePart type="family">Hinckeldeyn</namePart>
    <namePart type="given">Johannes</namePart>
  </name>
  <name type="personal">
    <namePart type="family">Kreutzfeldt</namePart>
    <namePart type="given">Jochen</namePart>
  </name>
  <abstract>Estimating the pose of a pallet and other logistics objects is crucial for various use cases, such as automatized material handling or tracking. Innovations in computer vision, computing power, and machine learning open up new opportunities for device-free localization based on cameras and neural networks. Large image datasets with annotated poses are required for training the network. Manual annotation, especially of 6D poses, is an extremely labor-intensive process. Hence, newer approaches often leverage synthetic training data to automatize the process of generating annotated image datasets. In this work, the generation of synthetic training data for 6D pose estimation of pallets is presented. The data is then used to train the Deep Object Pose Estimation (DOPE) algorithm [1]. The experimental validation of the algorithm proves that the 6D pose estimation of a standardized Euro pallet with a Red-Green-Blue (RGB) camera is feasible. The comparison of the results from three varying datasets under different lighting conditions shows the relevance of an appropriate dataset design to achieve an accurate and robust localization. The quantitative evaluation shows an average position error of less than 20 cm for the preferred dataset. The validated training dataset and a photorealistic model of a Euro pallet are publicly provided [2].</abstract>
  <subject>
    <topic>6D pose estimation</topic>
    <topic>6D-Posenschätzung</topic>
    <topic>DOPE algorithm</topic>
    <topic>DOPE-Algorithmus</topic>
    <topic>Europalette</topic>
    <topic>RGB camera</topic>
    <topic>RGB-Kamera</topic>
    <topic>Synthetischer Trainingsdatensatz</topic>
    <topic>euro pallet</topic>
    <topic>synthetic training dataset</topic>
  </subject>
  <classification authority="ddc">620</classification>
  <relatedItem type="host">
    <genre authority="marcgt">periodical</genre>
    <genre>academic journal</genre>
    <titleInfo>
      <title>Logistics Journal : Proceedings</title>
    </titleInfo>
    <part>
      <detail type="volume">
        <number>2022</number>
      </detail>
      <detail type="issue">
        <number>18</number>
      </detail>
      <date>2022</date>
    </part>
  </relatedItem>
  <identifier type="issn">2192-9084</identifier>
  <identifier type="urn">urn:nbn:de:0009-14-55900</identifier>
  <identifier type="doi">10.2195/lj_proc_knitt_en_202211_01</identifier>
  <identifier type="uri">http://nbn-resolving.de/urn:nbn:de:0009-14-55900</identifier>
  <identifier type="citekey">knitt2022</identifier>
</mods>
Download

Full Metadata

Bibliographic Citation Logistics Journal : referierte Veröffentlichungen, Vol. 2022, Iss. 18
Title

Estimating the Pose of a Euro Pallet with an RGB Camera based on Synthetic Training Data (eng)
Author Markus Knitt, Jakob Schyga, Asan Adamanov, Johannes Hinckeldeyn, Jochen Kreutzfeldt
Language eng
Abstract Estimating the pose of a pallet and other logistics objects is crucial for various use cases, such as automatized material handling or tracking. Innovations in computer vision, computing power, and machine learning open up new opportunities for device-free localization based on cameras and neural networks. Large image datasets with annotated poses are required for training the network. Manual annotation, especially of 6D poses, is an extremely labor-intensive process. Hence, newer approaches often leverage synthetic training data to automatize the process of generating annotated image datasets. In this work, the generation of synthetic training data for 6D pose estimation of pallets is presented. The data is then used to train the Deep Object Pose Estimation (DOPE) algorithm [1]. The experimental validation of the algorithm proves that the 6D pose estimation of a standardized Euro pallet with a Red-Green-Blue (RGB) camera is feasible. The comparison of the results from three varying datasets under different lighting conditions shows the relevance of an appropriate dataset design to achieve an accurate and robust localization. The quantitative evaluation shows an average position error of less than 20 cm for the preferred dataset. The validated training dataset and a photorealistic model of a Euro pallet are publicly provided [2]. Posenschätzung einer Palette und anderer Logistikobjekte ist von entscheidender Bedeutung für verschiedene Anwendungsfälle, wie automatisiertes Handling oder Tracking. Innovationen in der Bilderkennung, Rechenleistung und maschinellem Lernen eröffnen kamerabasierten Ansätzen auf Basis neuronaler Netze neue Möglichkeiten für die gerätelose Lokalisierung. Hierfür werden große Trainingsdatensätze mit annotierten Posen benötigt. Die manuelle Annotation, insbesondere von 6D-Posen, ist ein äußerst arbeitsintensiver Prozess, weshalb neuere Ansätze oftmals auf synthetischen Trainingsdaten basieren. In dieser Arbeit wird die Generierung synthetischer Trainingsdaten für die 6D-Posenschätzung von Paletten vorgestellt. Anschließend werden die Daten verwendet, um den Deep Object Pose Estimation (DOPE)-Algorithmus [1] zu trainieren. Die ex¬perimentelle Validierung des Algorithmus belegt, dass die 6D-Posenschätzung einer Europalette mit einer Rot-Grün-Blau (RGB) Kamera möglich ist. Der Vergleich der Ergebnisse von drei variierenden Datensätzen unter verschiedenen Lichtverhältnissen zeigt die Relevanz eines geeigneten Datensatzdesigns, um eine genaue und robuste Lokalisierung zu erreichen. Die quantitative Auswertung zeigt für den bevorzugten Datensatz einen durchschnittlichen Positionsfehler von weniger als 20 cm. Der validierte Trainingsdatensatz und ein fotorealistisches Modell einer Europalette sind öffentlich zur Verfügung gestellt [2].
Subject 6D pose estimation, 6D-Posenschätzung, DOPE algorithm, DOPE-Algorithmus, Europalette, RGB camera, RGB-Kamera, Synthetischer Trainingsdatensatz, euro pallet, synthetic training dataset
DDC 620
Rights fDPPL
URN: urn:nbn:de:0009-14-55900
DOI https://doi.org/10.2195/lj_proc_knitt_en_202211_01