Matches in SemOpenAlex for { <https://semopenalex.org/work/W3034973368> ?p ?o ?g. }
Showing items 1 to 93 of
93
with 100 items per page.
- W3034973368 abstract "Deformable convolutional networks have demonstrated outstanding performance in object recognition tasks with an effective feature extraction. Unlike standard convolution, the deformable convolution decides the receptive field size using dynamically generated offsets, which leads to an irregular memory access. Especially, the memory access pattern varies both spatially and temporally, making static optimization ineffective. Thus, a naive implementation would lead to an excessive memory footprint. In this paper, we present a novel approach to accelerate deformable convolution on FPGA. First, we propose a novel training method to reduce the size of the receptive field in the deformable convolutional layer without compromising accuracy. By optimizing the receptive field, we can compress the maximum size of the receptive field by 12.6 times. Second, we propose an efficient systolic architecture to maximize its efficiency. We then implement our design on FPGA to support the optimized dataflow. Experimental results show that our accelerator achieves up to 17.25 times speedup over the state-of-the-art accelerator." @default.
- W3034973368 created "2020-06-19" @default.
- W3034973368 creator A5032639277 @default.
- W3034973368 creator A5047681509 @default.
- W3034973368 creator A5084904773 @default.
- W3034973368 date "2020-06-09" @default.
- W3034973368 modified "2023-09-25" @default.
- W3034973368 title "An Efficient Accelerator Design Methodology for Deformable Convolutional Networks" @default.
- W3034973368 cites W1515422725 @default.
- W3034973368 cites W1686810756 @default.
- W3034973368 cites W1861492603 @default.
- W3034973368 cites W2094756095 @default.
- W3034973368 cites W2102605133 @default.
- W3034973368 cites W2163605009 @default.
- W3034973368 cites W2194775991 @default.
- W3034973368 cites W2276486856 @default.
- W3034973368 cites W2549139847 @default.
- W3034973368 cites W2566075159 @default.
- W3034973368 cites W2625954420 @default.
- W3034973368 cites W2944165510 @default.
- W3034973368 cites W2950141105 @default.
- W3034973368 cites W2950477723 @default.
- W3034973368 cites W2951548327 @default.
- W3034973368 cites W2953106684 @default.
- W3034973368 cites W3013525232 @default.
- W3034973368 cites W3023071679 @default.
- W3034973368 doi "https://doi.org/10.48550/arxiv.2006.05238" @default.
- W3034973368 hasPublicationYear "2020" @default.
- W3034973368 type Work @default.
- W3034973368 sameAs 3034973368 @default.
- W3034973368 citedByCount "0" @default.
- W3034973368 crossrefType "posted-content" @default.
- W3034973368 hasAuthorship W3034973368A5032639277 @default.
- W3034973368 hasAuthorship W3034973368A5047681509 @default.
- W3034973368 hasAuthorship W3034973368A5084904773 @default.
- W3034973368 hasBestOaLocation W30349733681 @default.
- W3034973368 hasConcept C111919701 @default.
- W3034973368 hasConcept C113775141 @default.
- W3034973368 hasConcept C138885662 @default.
- W3034973368 hasConcept C154945302 @default.
- W3034973368 hasConcept C173608175 @default.
- W3034973368 hasConcept C19071747 @default.
- W3034973368 hasConcept C202444582 @default.
- W3034973368 hasConcept C2776401178 @default.
- W3034973368 hasConcept C33923547 @default.
- W3034973368 hasConcept C41008148 @default.
- W3034973368 hasConcept C41895202 @default.
- W3034973368 hasConcept C42935608 @default.
- W3034973368 hasConcept C45347329 @default.
- W3034973368 hasConcept C50644808 @default.
- W3034973368 hasConcept C68339613 @default.
- W3034973368 hasConcept C74912251 @default.
- W3034973368 hasConcept C81363708 @default.
- W3034973368 hasConcept C9390403 @default.
- W3034973368 hasConcept C96324660 @default.
- W3034973368 hasConcept C9652623 @default.
- W3034973368 hasConceptScore W3034973368C111919701 @default.
- W3034973368 hasConceptScore W3034973368C113775141 @default.
- W3034973368 hasConceptScore W3034973368C138885662 @default.
- W3034973368 hasConceptScore W3034973368C154945302 @default.
- W3034973368 hasConceptScore W3034973368C173608175 @default.
- W3034973368 hasConceptScore W3034973368C19071747 @default.
- W3034973368 hasConceptScore W3034973368C202444582 @default.
- W3034973368 hasConceptScore W3034973368C2776401178 @default.
- W3034973368 hasConceptScore W3034973368C33923547 @default.
- W3034973368 hasConceptScore W3034973368C41008148 @default.
- W3034973368 hasConceptScore W3034973368C41895202 @default.
- W3034973368 hasConceptScore W3034973368C42935608 @default.
- W3034973368 hasConceptScore W3034973368C45347329 @default.
- W3034973368 hasConceptScore W3034973368C50644808 @default.
- W3034973368 hasConceptScore W3034973368C68339613 @default.
- W3034973368 hasConceptScore W3034973368C74912251 @default.
- W3034973368 hasConceptScore W3034973368C81363708 @default.
- W3034973368 hasConceptScore W3034973368C9390403 @default.
- W3034973368 hasConceptScore W3034973368C96324660 @default.
- W3034973368 hasConceptScore W3034973368C9652623 @default.
- W3034973368 hasLocation W30349733681 @default.
- W3034973368 hasOpenAccess W3034973368 @default.
- W3034973368 hasPrimaryLocation W30349733681 @default.
- W3034973368 hasRelatedWork W184060744 @default.
- W3034973368 hasRelatedWork W2335592298 @default.
- W3034973368 hasRelatedWork W2535649113 @default.
- W3034973368 hasRelatedWork W2770717529 @default.
- W3034973368 hasRelatedWork W2949649078 @default.
- W3034973368 hasRelatedWork W3013306997 @default.
- W3034973368 hasRelatedWork W3034868431 @default.
- W3034973368 hasRelatedWork W3034973368 @default.
- W3034973368 hasRelatedWork W3089673885 @default.
- W3034973368 hasRelatedWork W3092357178 @default.
- W3034973368 isParatext "false" @default.
- W3034973368 isRetracted "false" @default.
- W3034973368 magId "3034973368" @default.
- W3034973368 workType "article" @default.