Logo Detection With No Priors

In recent years, top referred methods on object detection like R-CNN have implemented this task as a combination of proposal region generation and supervised classification on the proposed bounding boxes. Although this pipeline has achieved state-of-the-art results in multiple datasets, it has inher...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Diego A. Velazquez, Josep M. Gonfaus, Pau Rodriguez, F. Xavier Roca, Seiichi Ozawa, Jordi Gonzalez
Formato:	article
Lenguaje:	EN
Publicado:	IEEE 2021
Materias:	Object detection transformers logo detection deep learning attention Electrical engineering. Electronics. Nuclear engineering TK1-9971
Acceso en línea:	https://doaj.org/article/06053aaa1432435eaf1b7fc745bef20f
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:06053aaa1432435eaf1b7fc745bef20f
record_format	dspace
spelling	oai:doaj.org-article:06053aaa1432435eaf1b7fc745bef20f2021-12-01T00:01:40ZLogo Detection With No Priors2169-353610.1109/ACCESS.2021.3101297https://doaj.org/article/06053aaa1432435eaf1b7fc745bef20f2021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9502074/https://doaj.org/toc/2169-3536In recent years, top referred methods on object detection like R-CNN have implemented this task as a combination of proposal region generation and supervised classification on the proposed bounding boxes. Although this pipeline has achieved state-of-the-art results in multiple datasets, it has inherent limitations that make object detection a very complex and inefficient task in computational terms. Instead of considering this standard strategy, in this paper we enhance Detection Transformers (DETR) which tackles object detection as a set-prediction problem directly in an end-to-end fully differentiable pipeline without requiring priors. In particular, we incorporate Feature Pyramids (FP) to the DETR architecture and demonstrate the effectiveness of the resulting DETR-FP approach on improving logo detection results thanks to the improved detection of small logos. So, without requiring any domain specific prior to be fed to the model, DETR-FP obtains competitive results on the OpenLogo and MS-COCO datasets offering a relative improvement of up to 30%, when compared to a Faster R-CNN baseline which strongly depends on hand-designed priors.Diego A. VelazquezJosep M. GonfausPau RodriguezF. Xavier RocaSeiichi OzawaJordi GonzalezIEEEarticleObject detectiontransformerslogo detectiondeep learningattentionElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENIEEE Access, Vol 9, Pp 106998-107011 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Object detection transformers logo detection deep learning attention Electrical engineering. Electronics. Nuclear engineering TK1-9971
spellingShingle	Object detection transformers logo detection deep learning attention Electrical engineering. Electronics. Nuclear engineering TK1-9971 Diego A. Velazquez Josep M. Gonfaus Pau Rodriguez F. Xavier Roca Seiichi Ozawa Jordi Gonzalez Logo Detection With No Priors
description	In recent years, top referred methods on object detection like R-CNN have implemented this task as a combination of proposal region generation and supervised classification on the proposed bounding boxes. Although this pipeline has achieved state-of-the-art results in multiple datasets, it has inherent limitations that make object detection a very complex and inefficient task in computational terms. Instead of considering this standard strategy, in this paper we enhance Detection Transformers (DETR) which tackles object detection as a set-prediction problem directly in an end-to-end fully differentiable pipeline without requiring priors. In particular, we incorporate Feature Pyramids (FP) to the DETR architecture and demonstrate the effectiveness of the resulting DETR-FP approach on improving logo detection results thanks to the improved detection of small logos. So, without requiring any domain specific prior to be fed to the model, DETR-FP obtains competitive results on the OpenLogo and MS-COCO datasets offering a relative improvement of up to 30%, when compared to a Faster R-CNN baseline which strongly depends on hand-designed priors.
format	article
author	Diego A. Velazquez Josep M. Gonfaus Pau Rodriguez F. Xavier Roca Seiichi Ozawa Jordi Gonzalez
author_facet	Diego A. Velazquez Josep M. Gonfaus Pau Rodriguez F. Xavier Roca Seiichi Ozawa Jordi Gonzalez
author_sort	Diego A. Velazquez
title	Logo Detection With No Priors
title_short	Logo Detection With No Priors
title_full	Logo Detection With No Priors
title_fullStr	Logo Detection With No Priors
title_full_unstemmed	Logo Detection With No Priors
title_sort	logo detection with no priors
publisher	IEEE
publishDate	2021
url	https://doaj.org/article/06053aaa1432435eaf1b7fc745bef20f
work_keys_str_mv	AT diegoavelazquez logodetectionwithnopriors AT josepmgonfaus logodetectionwithnopriors AT paurodriguez logodetectionwithnopriors AT fxavierroca logodetectionwithnopriors AT seiichiozawa logodetectionwithnopriors AT jordigonzalez logodetectionwithnopriors
_version_	1718406122601185280

Logo Detection With No Priors

Ejemplares similares