File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Conference Paper: Inception Convolution with Efficient Dilation Search

TitleInception Convolution with Efficient Dilation Search
Authors
Issue Date2021
Citation
Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2021, p. 11481-11490 How to Cite?
AbstractAs a variant of standard convolution, a dilated convolution can control effective receptive fields and handle large scale variance of objects without introducing additional computational costs. To fully explore the potential of dilated convolution, we proposed a new type of dilated convolution (referred to as inception convolution), where the convolution operations have independent dilation patterns among different axes, channels and layers. To develop a practical method for learning complex inception convolution based on the data, a simple but effective search algorithm, referred to as efficient dilation optimization (EDO), is developed. Based on statistical optimization, the EDO method operates in a low-cost manner and is extremely fast when it is applied on large scale datasets. Empirical results validate that our method achieves consistent performance gains for image recognition, object detection, instance segmentation, human detection, and human pose estimation. For instance, by simply replacing the 3 × 3 standard convolution in the ResNet-50 backbone with inception convolution, we significantly improve the AP of Faster R-CNN from 36.4% to 39.2% on MS COCO.
Persistent Identifierhttp://hdl.handle.net/10722/321962
ISSN
2023 SCImago Journal Rankings: 10.331
ISI Accession Number ID

 

DC FieldValueLanguage
dc.contributor.authorLiu, Jie-
dc.contributor.authorLi, Chuming-
dc.contributor.authorLiang, Feng-
dc.contributor.authorLin, Chen-
dc.contributor.authorSun, Ming-
dc.contributor.authorYan, Junjie-
dc.contributor.authorOuyang, Wanli-
dc.contributor.authorXu, Dong-
dc.date.accessioned2022-11-03T02:22:39Z-
dc.date.available2022-11-03T02:22:39Z-
dc.date.issued2021-
dc.identifier.citationProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2021, p. 11481-11490-
dc.identifier.issn1063-6919-
dc.identifier.urihttp://hdl.handle.net/10722/321962-
dc.description.abstractAs a variant of standard convolution, a dilated convolution can control effective receptive fields and handle large scale variance of objects without introducing additional computational costs. To fully explore the potential of dilated convolution, we proposed a new type of dilated convolution (referred to as inception convolution), where the convolution operations have independent dilation patterns among different axes, channels and layers. To develop a practical method for learning complex inception convolution based on the data, a simple but effective search algorithm, referred to as efficient dilation optimization (EDO), is developed. Based on statistical optimization, the EDO method operates in a low-cost manner and is extremely fast when it is applied on large scale datasets. Empirical results validate that our method achieves consistent performance gains for image recognition, object detection, instance segmentation, human detection, and human pose estimation. For instance, by simply replacing the 3 × 3 standard convolution in the ResNet-50 backbone with inception convolution, we significantly improve the AP of Faster R-CNN from 36.4% to 39.2% on MS COCO.-
dc.languageeng-
dc.relation.ispartofProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition-
dc.titleInception Convolution with Efficient Dilation Search-
dc.typeConference_Paper-
dc.description.naturelink_to_subscribed_fulltext-
dc.identifier.doi10.1109/CVPR46437.2021.01132-
dc.identifier.scopuseid_2-s2.0-85113838663-
dc.identifier.spage11481-
dc.identifier.epage11490-
dc.identifier.isiWOS:000742075001066-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats