Show simple item record

dc.contributor.authorOro García, David
dc.contributor.authorFernandez Tena, Carles
dc.contributor.authorMartorell Bofill, Xavier
dc.contributor.authorHernando Pericás, Francisco Javier
dc.contributor.otherUniversitat Politècnica de Catalunya. Doctorat en Teoria del Senyal i Comunicacions
dc.contributor.otherUniversitat Politècnica de Catalunya. Departament d'Arquitectura de Computadors
dc.contributor.otherUniversitat Politècnica de Catalunya. Departament de Teoria del Senyal i Comunicacions
dc.date.accessioned2021-01-26T11:18:20Z
dc.date.available2021-08-21T00:32:34Z
dc.date.issued2020-08-21
dc.identifier.citationOro, D. [et al.]. Work-efficient parallel non-maximum suppression kernels. "Computer journal", 21 Agost 2020, núm. bxaa108, p. 1-15.
dc.identifier.issn0010-4620
dc.identifier.urihttp://hdl.handle.net/2117/335999
dc.description.abstractIn the context of object detection, sliding-window classifiers and single-shot convolutional neural network (CNN) meta-architectures typically yield multiple overlapping candidate windows with similar high scores around the true location of a particular object. Non-maximum suppression (NMS) is the process of selecting a single representative candidate within this cluster of detections, so as to obtain a unique detection per object appearing on a given picture. In this paper, we present a highly scalable NMS algorithm for embedded graphics processing unit (GPU) architectures that is designed from scratch to handle workloads featuring thousands of simultaneous detections on a given picture. Our kernels are directly applicable to other sequential NMS algorithms such as FeatureNMS, Soft-NMS or AdaptiveNMS that share the inner workings of the classic greedy NMS method. The obtained performance results show that our parallel NMS algorithm is capable of clustering 1024 simultaneous detected objects per frame in roughly 1 ms on both Tegra X1 and Tegra X2 on-die GPUs, while taking 2 ms on Tegra K1. Furthermore, our proposed parallel greedy NMS algorithm yields a 14–40x speed up when compared to state-of-the-art NMS methods that require learning a CNN from annotated data.
dc.description.sponsorshipThis work has been partially supported by the Ministerio de Economía y Competitividad under contracts (TIN2015-65316-P, TEC2012-38939-C03-02), the Departament d’Innovació, Universitats i Empresa de la Generalitat de Catalunya under project MPEXPAR: Models de Programació i Entorns d’Execució Paral·lels (2014-SGR-1051), and the European Commission under the Horizon 2020 program (H2020-ICT-644312).
dc.format.extent15 p.
dc.language.isoeng
dc.publisherWiley Heyden
dc.subjectÀrees temàtiques de la UPC::Enginyeria de la telecomunicació::Processament del senyal::Processament de la imatge i del senyal vídeo
dc.subject.lcshImage processing
dc.subject.otherNon-maximum suppression
dc.subject.otherObject detection
dc.subject.otherGPU computing
dc.subject.otherParallel computing
dc.titleWork-efficient parallel non-maximum suppression kernels
dc.typeArticle
dc.subject.lemacImatges -- Processament
dc.contributor.groupUniversitat Politècnica de Catalunya. CAP - Grup de Computació d'Altes Prestacions
dc.contributor.groupUniversitat Politècnica de Catalunya. VEU - Grup de Tractament de la Parla
dc.identifier.doi10.1093/comjnl/bxaa108
dc.description.peerreviewedPeer Reviewed
dc.relation.publisherversionhttps://academic.oup.com/comjnl/advance-article-abstract/doi/10.1093/comjnl/bxaa108/5894719
dc.rights.accessOpen Access
local.identifier.drac30252151
dc.description.versionPostprint (author's final draft)
dc.relation.projectidinfo:eu-repo/grantAgreement/MINECO/6PN/TEC2012-38939-C03-02
dc.relation.projectidinfo:eu-repo/grantAgreement/MINECO/1PE/TIN2015-65316-P
dc.relation.projectidinfo:eu-repo/grantAgreement/AGAUR/V PRI/2014 SGR 1051
local.citation.authorOro, D.; Fernández, C.; Martorell, X.; Hernando, J.
local.citation.publicationNameComputer journal
local.citation.numberbxaa108
local.citation.startingPage1
local.citation.endingPage15


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record

All rights reserved. This work is protected by the corresponding intellectual and industrial property rights. Without prejudice to any existing legal exemptions, reproduction, distribution, public communication or transformation of this work are prohibited without permission of the copyright holder