Show simple item record

dc.contributor.authorHu, Ping
dc.contributor.authorPerazzi, Federico
dc.contributor.authorHeilbron, Fabian Caba
dc.contributor.authorWang, Oliver
dc.contributor.authorLin, Zhe
dc.contributor.authorSaenko, Kate
dc.contributor.authorSclaroff, Stan
dc.date.accessioned2020-11-22T12:31:01Z
dc.date.available2020-11-22T12:31:01Z
dc.date.issued2020-11-20
dc.date.submitted2020-08-25
dc.identifier.citationHu, P., Perazzi, F., Heilbron, F. C., Wang, O., Lin, Z., Saenko, K., & Sclaroff, S. (2020). Real-time Semantic Segmentation with Fast Attention. IEEE Robotics and Automation Letters, 1–1. doi:10.1109/lra.2020.3039744
dc.identifier.issn2377-3774
dc.identifier.doi10.1109/LRA.2020.3039744
dc.identifier.urihttp://hdl.handle.net/10754/666065
dc.description.abstractAccurate semantic segmentation requires rich contextual cues (large receptive fields) and fine spatial details (high resolution), both of which incur high computational costs. In this paper, we propose a novel architecture that addresses both challenges and achieves state-of-the-art performance for semantic segmentation of high-resolution images and videos in real-time. The proposed architecture relies on our fast attention, which is a simple modification of the popular self-attention mechanism and captures the same rich contextual information at a small fraction of the computational cost, by changing the order of operations. Moreover, to efficiently process high-resolution input, we apply an additional spatial reduction to intermediate feature stages of the network with minimal loss in accuracy thanks to the use of the fast attention module to fuse features. We validate our method with a series of experiments, and show that results on multiple datasets demonstrate superior performance with better accuracy and speed compared to existing approaches for real-time semantic segmentation. On Cityscapes, our network achieves 74.4% mIoU at 72 FPS and 75.5% mIoU at 58 FPS on a single Titan X GPU, which is ~50% faster than the state-of-the-art while retaining the same accuracy.
dc.publisherIEEE
dc.relation.urlhttps://ieeexplore.ieee.org/document/9265219/
dc.relation.urlhttps://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9265219
dc.relation.urlhttp://arxiv.org/pdf/2007.03815
dc.rights(c) 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.
dc.rightsThis file is an open access version redistributed from: http://arxiv.org/pdf/2007.03815
dc.subjectSemantic Segmentation
dc.subjectReal-time Speed
dc.subjectFast Attention
dc.titleReal-time Semantic Segmentation with Fast Attention
dc.typeArticle
dc.contributor.departmentKing Abdullah University of Science and Technology, KAUST, Saudi Arabia.
dc.identifier.journalIEEE Robotics and Automation Letters
dc.eprint.versionPre-print
dc.contributor.institutionComputer Science, Boston University, Boston, MA, United States of America, 02215
dc.contributor.institutionFacebook, United States of America,
dc.contributor.institutionAdobe, United States of America,
dc.contributor.institutionAdobe Research, Adobe Research, SEATTLE, WA, United States of America, 98103
dc.contributor.institutionAdobe Research, Adobe Systems, Inc., Fremont, CA, United States of America, 94539
dc.contributor.institutionCS, Boston University, Boston, Massachusetts, United States of America, 02215
dc.contributor.institutionBoston University, United States of America,
dc.identifier.pages1-1
dc.identifier.arxivid2007.03815
kaust.personHeilbron, Fabian Caba
dc.date.accepted2020-09-13
refterms.dateFOA2021-06-28T13:36:01Z
dc.date.published-online2020-11-20
dc.date.published-print2021-01


Files in this item

Thumbnail
Name:
Articlefile1.pdf
Size:
2.001Mb
Format:
PDF
Description:
Pre-print

This item appears in the following Collection(s)

Show simple item record