Show simple item record

dc.contributor.authorMerino, Ibon
dc.contributor.authorAzpiazu, Jon
dc.contributor.authorRemazeilles, Anthony
dc.contributor.authorSierra, Basilio
dc.date.accessioned2021-02-09T10:52:53Z
dc.date.available2021-02-09T10:52:53Z
dc.date.issued2021-02-04
dc.identifier.citationMerino, Ibon, Jon Azpiazu, Anthony Remazeilles, and Basilio Sierra. “3D Convolutional Neural Networks Initialized from Pretrained 2D Convolutional Neural Networks for Classification of Industrial Parts.” Sensors 21, no. 4 (February 4, 2021): 1078. doi:10.3390/s21041078.en
dc.identifier.urihttp://hdl.handle.net/11556/1077
dc.description.abstractDeep learning methods have been successfully applied to image processing, mainly using 2D vision sensors. Recently, the rise of depth cameras and other similar 3D sensors has opened the field for new perception techniques. Nevertheless, 3D convolutional neural networks perform slightly worse than other 3D deep learning methods, and even worse than their 2D version. In this paper, we propose to improve 3D deep learning results by transferring the pretrained weights learned in 2D networks to their corresponding 3D version. Using an industrial object recognition context, we have analyzed different combinations of 3D convolutional networks (VGG16, ResNet, Inception ResNet, and EfficientNet), comparing the recognition accuracy. The highest accuracy is obtained with EfficientNetB0 using extrusion with an accuracy of 0.9217, which gives comparable results to state-of-the art methods. We also observed that the transfer approach enabled to improve the accuracy of the Inception ResNet 3D version up to 18% with respect to the score of the 3D approach alone.en
dc.description.sponsorshipThis paper has been supported by the project ELKARBOT under the Basque program ELKARTEK, grant agreement No. KK-2020/00092.en
dc.language.isoengen
dc.publisherMultidisciplinary Digital Publishing Institute (MDPI)en
dc.rightsAttribution 4.0 International*
dc.rights.urihttp://creativecommons.org/licenses/by/4.0/*
dc.title3D Convolutional Neural Networks Initialized from Pretrained 2D Convolutional Neural Networks for Classification of Industrial Partsen
dc.typearticleen
dc.identifier.doi10.3390/s21041078en
dc.rights.accessRightsopenAccessen
dc.subject.keywordsComputer visionen
dc.subject.keywordsDeep learningen
dc.subject.keywordsTransfer learningen
dc.subject.keywordsObject recognitionen
dc.identifier.essn1424-8220en
dc.issue.number4en
dc.journal.titleSensorsen
dc.page.initial1078en
dc.volume.number21en


Files in this item

Thumbnail

    Show simple item record

    Attribution 4.0 InternationalExcept where otherwise noted, this item's license is described as Attribution 4.0 International