Matches in SemOpenAlex for { <https://semopenalex.org/work/W4387447991> ?p ?o ?g. }
Showing items 1 to 83 of
83
with 100 items per page.
- W4387447991 endingPage "14" @default.
- W4387447991 startingPage "1" @default.
- W4387447991 abstract "In recent years, natural, flexible, and contactless vision-based gesture recognition has received significant attention in human-computer interaction. However, employing convolutional neural networks (CNNs) for RGB or RGB-D gestures can result in excessive power consumption and poor energy efficiency, making them unsuitable for embedded systems. In this paper, we propose a lightweight sparse binarized neural network (sBNN) model for edge gesture recognition that achieves an accuracy of 89.43%-99.92% on four open-source gesture datasets with <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$leq$</tex-math> </inline-formula> 20.26 million operations (MOP) and <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$leq$</tex-math> </inline-formula> 15.83-Kilobytes (KB) parameters. We find high channel-level sparsity in the activation maps of sBNN when edge gestures are used as inputs. The sparse activation maps have multiple identical activation vectors called sparse activation vectors (SAV), which lead to highly repeated calculations. In order to avoid this issue, we propose a two-stage value prediction approach to skip these calculations, achieving a speedup of 1.03x-1.83x. Moreover, to reduce on-chip memory, the compression technique is applied to the sparse activation maps, providing a compression rate of 1.72x-3.45x. Finally, we implement an energy-efficient sparse BNN accelerator (SBA) on an embedded field-programmable gate array (FPGA). The experimental results show that our SBA has a latency of 26.3-46.8- <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$mu$</tex-math> </inline-formula> s, a power consumption of 0.807 W, and an energy efficiency of 536.22-952.70-GOPS/W at 50-MHz frequency. Our SBA offers lower latency, lower power consumption, and higher energy efficiency than previous state-of-the-art gesture recognition accelerators." @default.
- W4387447991 created "2023-10-10" @default.
- W4387447991 creator A5013065799 @default.
- W4387447991 creator A5022802322 @default.
- W4387447991 creator A5028407427 @default.
- W4387447991 creator A5049428400 @default.
- W4387447991 creator A5049845476 @default.
- W4387447991 creator A5053834265 @default.
- W4387447991 creator A5066674438 @default.
- W4387447991 creator A5075507579 @default.
- W4387447991 creator A5089781655 @default.
- W4387447991 date "2023-01-01" @default.
- W4387447991 modified "2023-10-11" @default.
- W4387447991 title "An Energy-Efficient BNN Accelerator With Two-Stage Value Prediction for Sparse-Edge Gesture Recognition" @default.
- W4387447991 doi "https://doi.org/10.1109/tcsi.2023.3320175" @default.
- W4387447991 hasPublicationYear "2023" @default.
- W4387447991 type Work @default.
- W4387447991 citedByCount "0" @default.
- W4387447991 crossrefType "journal-article" @default.
- W4387447991 hasAuthorship W4387447991A5013065799 @default.
- W4387447991 hasAuthorship W4387447991A5022802322 @default.
- W4387447991 hasAuthorship W4387447991A5028407427 @default.
- W4387447991 hasAuthorship W4387447991A5049428400 @default.
- W4387447991 hasAuthorship W4387447991A5049845476 @default.
- W4387447991 hasAuthorship W4387447991A5053834265 @default.
- W4387447991 hasAuthorship W4387447991A5066674438 @default.
- W4387447991 hasAuthorship W4387447991A5075507579 @default.
- W4387447991 hasAuthorship W4387447991A5089781655 @default.
- W4387447991 hasConcept C105795698 @default.
- W4387447991 hasConcept C111919701 @default.
- W4387447991 hasConcept C11413529 @default.
- W4387447991 hasConcept C138236772 @default.
- W4387447991 hasConcept C149635348 @default.
- W4387447991 hasConcept C154945302 @default.
- W4387447991 hasConcept C159437735 @default.
- W4387447991 hasConcept C162307627 @default.
- W4387447991 hasConcept C173608175 @default.
- W4387447991 hasConcept C186370098 @default.
- W4387447991 hasConcept C207347870 @default.
- W4387447991 hasConcept C33923547 @default.
- W4387447991 hasConcept C41008148 @default.
- W4387447991 hasConcept C42935608 @default.
- W4387447991 hasConcept C68339613 @default.
- W4387447991 hasConcept C79974875 @default.
- W4387447991 hasConcept C81363708 @default.
- W4387447991 hasConcept C82990744 @default.
- W4387447991 hasConceptScore W4387447991C105795698 @default.
- W4387447991 hasConceptScore W4387447991C111919701 @default.
- W4387447991 hasConceptScore W4387447991C11413529 @default.
- W4387447991 hasConceptScore W4387447991C138236772 @default.
- W4387447991 hasConceptScore W4387447991C149635348 @default.
- W4387447991 hasConceptScore W4387447991C154945302 @default.
- W4387447991 hasConceptScore W4387447991C159437735 @default.
- W4387447991 hasConceptScore W4387447991C162307627 @default.
- W4387447991 hasConceptScore W4387447991C173608175 @default.
- W4387447991 hasConceptScore W4387447991C186370098 @default.
- W4387447991 hasConceptScore W4387447991C207347870 @default.
- W4387447991 hasConceptScore W4387447991C33923547 @default.
- W4387447991 hasConceptScore W4387447991C41008148 @default.
- W4387447991 hasConceptScore W4387447991C42935608 @default.
- W4387447991 hasConceptScore W4387447991C68339613 @default.
- W4387447991 hasConceptScore W4387447991C79974875 @default.
- W4387447991 hasConceptScore W4387447991C81363708 @default.
- W4387447991 hasConceptScore W4387447991C82990744 @default.
- W4387447991 hasFunder F4320321001 @default.
- W4387447991 hasLocation W43874479911 @default.
- W4387447991 hasOpenAccess W4387447991 @default.
- W4387447991 hasPrimaryLocation W43874479911 @default.
- W4387447991 hasRelatedWork W2010878661 @default.
- W4387447991 hasRelatedWork W2026258298 @default.
- W4387447991 hasRelatedWork W2185750513 @default.
- W4387447991 hasRelatedWork W2805039731 @default.
- W4387447991 hasRelatedWork W2902873204 @default.
- W4387447991 hasRelatedWork W3147379364 @default.
- W4387447991 hasRelatedWork W3204639664 @default.
- W4387447991 hasRelatedWork W342863993 @default.
- W4387447991 hasRelatedWork W4312416068 @default.
- W4387447991 hasRelatedWork W2970836791 @default.
- W4387447991 isParatext "false" @default.
- W4387447991 isRetracted "false" @default.
- W4387447991 workType "article" @default.