Identity statement area
Reference TypeConference Proceedings
Last Update2020:
Metadata Last Update2020: administrator
Citation KeyBatistadaCunhaSanValMagTei:2020:StImDo
TitleA Study on the Impact of Domain Randomization for Monocular Deep 6DoF Pose Estimation
DateNov. 7-10, 2020
Access Date2020, Dec. 04
Number of Files1
Size9047 KiB
Context area
Author1 Batista da Cunha, Kelvin
2 dos Santos Brito, Caio José
3 Valença da Rocha Martins Albuquerque, Lucas
4 Magalhães Simões, Francisco Paulo
5 Teichrieb, Veronica
Affiliation1 Voxar Labs, Centro de Informática, Universidade Federal de Pernambuco
2 Voxar Labs, Centro de Informática, Universidade Federal de Pernambuco
3 Voxar Labs, Centro de Informática, Universidade Federal de Pernambuco
4 Curso Técnico em Informática para Internet, Instituto Federal de Pernambuco, Campus Belo Jardim
5 Voxar Labs, Centro de Informática, Universidade Federal de Pernambuco
EditorMusse, Soraia Raupp
Cesar Junior, Roberto Marcondes
Pelechano, Nuria
Wang, Zhangyang (Atlas)
Conference NameConference on Graphics, Patterns and Images, 33 (SIBGRAPI)
Conference LocationVirtual
Book TitleProceedings
PublisherIEEE Computer Society
Publisher CityLos Alamitos
History2020-09-30 16:45:08 :: -> administrator ::
2020-10-28 20:46:58 :: administrator -> :: 2020
Content and structure area
Is the master or a copy?is the master
Document Stagecompleted
Document Stagenot transferred
Content TypeExternal Contribution
Tertiary TypeFull Paper
KeywordsPose Estimation, Deep Learning, Domain Randomization.
AbstractIn this work, we apply domain randomization to synthetic images and train deep 6DoF monocular RGB pose estimation models to work on a real object. We compare 19 models trained with different combinations of synthetic and real data (fully synthetic, fully real, initially synthetic and supplemented with real, and a real-synthetic randomized mix). By gradually decreasing the amount of real data used, we show it is possible for deep 6DoF detection to obtain superior results while using less real data (which is harder to obtain) and suggest the best approach to train a model with synthetic data. Our method is validated using a textureless 3D printed object, as the textureless category is a challenging, common open problem in itself. A real and a synthetic dataset generated for this work, totalling over 24,800 annotated frames, are also made public. We also show that synthetic, randomized data can help generalize a model by training it to handle challenges such as illumination changes and fast motion. Finally, we also evaluate how a model trained for one camera sensor works with a different one, and show that synthetic simulations of real cameras can help overcoming this challenge.
source Directory Contentthere are no files
agreement Directory Content
agreement.html 30/09/2020 13:45 1.2 KiB 
Conditions of access and use area
Target File108.pdf
e-Mail (login)
Allied materials area
Next Higher Units8JMKD3MGPEW34M/43G4L9S
Notes area
Empty Fieldsaccessionnumber archivingpolicy archivist area callnumber copyholder copyright creatorhistory descriptionlevel dissemination doi edition electronicmailaddress group holdercode isbn issn label lineage mark nextedition notes numberofvolumes orcid organization pages parameterlist parentrepositories previousedition previouslowerunit progress project readergroup readpermission resumeid rightsholder secondarydate secondarykey secondarymark secondarytype serieseditor session shorttitle sponsor subject tertiarymark type url versiontype volume
Description control area