Matches in SemOpenAlex for { <https://semopenalex.org/work/W4312095862> ?p ?o ?g. }
- W4312095862 abstract "Due to the high parallelism and efficiency, Convolution acceleration and buffer pipeline architecture are common solutions for convolutional neural network (CNN) acceleration, however, existing accelerate solutions exhibit limited performance to accelerate CNN models. This paper presents a novel architecture as YOLO accelerate solution, which based on convolution loop unrolling and data flow pipeline optimization to reduce power consumption and maintain computational accuracy. The convolution loop unrolling algorithm is used to convert the numerous multiplication operations into addition operations to reduce the amount of computation. In the meanwhile, a buffer pipeline method which splits the processing element (PE) array is proposed, further improves the efficiency and optimizes the memory of the accelerator. Moreover, the data of the original model are processed at a low fixed point, reducing the resource consumption of the FPGA. Our proposed architecture can ensure the accuracy of network testing while reducing the power consumption required by the hardware." @default.
- W4312095862 created "2023-01-04" @default.
- W4312095862 creator A5033279662 @default.
- W4312095862 creator A5074254868 @default.
- W4312095862 creator A5075985115 @default.
- W4312095862 creator A5088144038 @default.
- W4312095862 date "2022-11-05" @default.
- W4312095862 modified "2023-09-30" @default.
- W4312095862 title "A power-efficient optimizing framework FPGA accelerator for YOLO" @default.
- W4312095862 cites W2276486856 @default.
- W4312095862 cites W2463631526 @default.
- W4312095862 cites W2513568085 @default.
- W4312095862 cites W2604233003 @default.
- W4312095862 cites W2605487586 @default.
- W4312095862 cites W2606722458 @default.
- W4312095862 cites W2729080111 @default.
- W4312095862 cites W2761085955 @default.
- W4312095862 cites W2777186991 @default.
- W4312095862 cites W2790167166 @default.
- W4312095862 cites W2806397907 @default.
- W4312095862 cites W2908675692 @default.
- W4312095862 cites W2935524202 @default.
- W4312095862 cites W2944090748 @default.
- W4312095862 cites W2944669570 @default.
- W4312095862 cites W2947350915 @default.
- W4312095862 cites W2954859667 @default.
- W4312095862 cites W2963460857 @default.
- W4312095862 cites W2963809521 @default.
- W4312095862 cites W2975291694 @default.
- W4312095862 cites W2980702327 @default.
- W4312095862 cites W2995942064 @default.
- W4312095862 cites W3014927913 @default.
- W4312095862 cites W3034100311 @default.
- W4312095862 cites W3083622693 @default.
- W4312095862 cites W3086928989 @default.
- W4312095862 cites W4253012315 @default.
- W4312095862 doi "https://doi.org/10.1109/cisp-bmei56279.2022.9980321" @default.
- W4312095862 hasPublicationYear "2022" @default.
- W4312095862 type Work @default.
- W4312095862 citedByCount "1" @default.
- W4312095862 countsByYear W43120958622023 @default.
- W4312095862 crossrefType "proceedings-article" @default.
- W4312095862 hasAuthorship W4312095862A5033279662 @default.
- W4312095862 hasAuthorship W4312095862A5074254868 @default.
- W4312095862 hasAuthorship W4312095862A5075985115 @default.
- W4312095862 hasAuthorship W4312095862A5088144038 @default.
- W4312095862 hasConcept C113775141 @default.
- W4312095862 hasConcept C118993495 @default.
- W4312095862 hasConcept C121332964 @default.
- W4312095862 hasConcept C127413603 @default.
- W4312095862 hasConcept C13164978 @default.
- W4312095862 hasConcept C149635348 @default.
- W4312095862 hasConcept C154945302 @default.
- W4312095862 hasConcept C163258240 @default.
- W4312095862 hasConcept C169590947 @default.
- W4312095862 hasConcept C173608175 @default.
- W4312095862 hasConcept C175309249 @default.
- W4312095862 hasConcept C199360897 @default.
- W4312095862 hasConcept C41008148 @default.
- W4312095862 hasConcept C42935608 @default.
- W4312095862 hasConcept C43521106 @default.
- W4312095862 hasConcept C45347329 @default.
- W4312095862 hasConcept C50644808 @default.
- W4312095862 hasConcept C62520636 @default.
- W4312095862 hasConcept C76970557 @default.
- W4312095862 hasConcept C81363708 @default.
- W4312095862 hasConcept C87717796 @default.
- W4312095862 hasConcept C9390403 @default.
- W4312095862 hasConceptScore W4312095862C113775141 @default.
- W4312095862 hasConceptScore W4312095862C118993495 @default.
- W4312095862 hasConceptScore W4312095862C121332964 @default.
- W4312095862 hasConceptScore W4312095862C127413603 @default.
- W4312095862 hasConceptScore W4312095862C13164978 @default.
- W4312095862 hasConceptScore W4312095862C149635348 @default.
- W4312095862 hasConceptScore W4312095862C154945302 @default.
- W4312095862 hasConceptScore W4312095862C163258240 @default.
- W4312095862 hasConceptScore W4312095862C169590947 @default.
- W4312095862 hasConceptScore W4312095862C173608175 @default.
- W4312095862 hasConceptScore W4312095862C175309249 @default.
- W4312095862 hasConceptScore W4312095862C199360897 @default.
- W4312095862 hasConceptScore W4312095862C41008148 @default.
- W4312095862 hasConceptScore W4312095862C42935608 @default.
- W4312095862 hasConceptScore W4312095862C43521106 @default.
- W4312095862 hasConceptScore W4312095862C45347329 @default.
- W4312095862 hasConceptScore W4312095862C50644808 @default.
- W4312095862 hasConceptScore W4312095862C62520636 @default.
- W4312095862 hasConceptScore W4312095862C76970557 @default.
- W4312095862 hasConceptScore W4312095862C81363708 @default.
- W4312095862 hasConceptScore W4312095862C87717796 @default.
- W4312095862 hasConceptScore W4312095862C9390403 @default.
- W4312095862 hasLocation W43120958621 @default.
- W4312095862 hasOpenAccess W4312095862 @default.
- W4312095862 hasPrimaryLocation W43120958621 @default.
- W4312095862 hasRelatedWork W2000117128 @default.
- W4312095862 hasRelatedWork W2770717529 @default.
- W4312095862 hasRelatedWork W2949649078 @default.
- W4312095862 hasRelatedWork W2978904333 @default.
- W4312095862 hasRelatedWork W3116557650 @default.
- W4312095862 hasRelatedWork W4287595677 @default.
- W4312095862 hasRelatedWork W4312081660 @default.