Matches in SemOpenAlex for { <https://semopenalex.org/work/W3048895265> ?p ?o ?g. }
Showing items 1 to 96 of
96
with 100 items per page.
- W3048895265 endingPage "400" @default.
- W3048895265 startingPage "388" @default.
- W3048895265 abstract "Suffering from both high computational complexity and high memory bandwidth is the major challenge in realizing the deep neural network in low-power for real-time applications. Binarizing the feature maps as well as the filter coefficients in deep neural network is an efficient way to reduce the high power consumption in deep learning object detection, however, it greatly scarifies the detection accuracy when reducing the bit-width of a 32-bit word to a binary bit in a floating-point deep neural network. This paper proposes a hybrid fixed point/binary deep neural network design methodology for object detection to achieve low-power consumption by taking advantage of both the fixed-point and binary deep neural networks, which allocates enough bit-width to design the hardware datapath in different layers of deep neural network. The proposed methodology combines dynamic fixed-point quantization and binarization techniques together to extremely compress the object detection model to result in a compact hybrid fixed-point/binary detection neural network, which achieves lower bandwidth and lower computational complexity. An automation tool based on the proposed methodology is also developed to train a hybrid deep neural network under a specified quality loss range. Taking MobileNet-SSD as an example, using the proposed methodology, the resulted model achieves 91% model size reduction and 75.8% memory bandwidth reduction at the cost of less than 1% mAP quality degradation. The proposed design methodology for hybrid fixed-point/binary deep neural networks achieves a good balance on detection accuracy, model size compression ratio and feature map reduction for low-power deep learning object detection applications." @default.
- W3048895265 created "2020-08-18" @default.
- W3048895265 creator A5007783236 @default.
- W3048895265 creator A5010546672 @default.
- W3048895265 creator A5022312926 @default.
- W3048895265 creator A5023187165 @default.
- W3048895265 creator A5064924455 @default.
- W3048895265 date "2020-09-01" @default.
- W3048895265 modified "2023-10-16" @default.
- W3048895265 title "Hybrid Fixed-Point/Binary Deep Neural Network Design Methodology for Low-Power Object Detection" @default.
- W3048895265 cites W1677182931 @default.
- W3048895265 cites W2031489346 @default.
- W3048895265 cites W2108598243 @default.
- W3048895265 cites W2194775991 @default.
- W3048895265 cites W2442974303 @default.
- W3048895265 cites W2554302513 @default.
- W3048895265 cites W2585560244 @default.
- W3048895265 cites W2586654419 @default.
- W3048895265 cites W2618939455 @default.
- W3048895265 cites W2750432752 @default.
- W3048895265 cites W2790167166 @default.
- W3048895265 cites W2793950911 @default.
- W3048895265 cites W2804514187 @default.
- W3048895265 cites W2898466789 @default.
- W3048895265 cites W2963122961 @default.
- W3048895265 cites W2968483100 @default.
- W3048895265 cites W2973657294 @default.
- W3048895265 cites W3140772298 @default.
- W3048895265 cites W4246550359 @default.
- W3048895265 doi "https://doi.org/10.1109/jetcas.2020.3015753" @default.
- W3048895265 hasPublicationYear "2020" @default.
- W3048895265 type Work @default.
- W3048895265 sameAs 3048895265 @default.
- W3048895265 citedByCount "6" @default.
- W3048895265 countsByYear W30488952652021 @default.
- W3048895265 countsByYear W30488952652022 @default.
- W3048895265 countsByYear W30488952652023 @default.
- W3048895265 crossrefType "journal-article" @default.
- W3048895265 hasAuthorship W3048895265A5007783236 @default.
- W3048895265 hasAuthorship W3048895265A5010546672 @default.
- W3048895265 hasAuthorship W3048895265A5022312926 @default.
- W3048895265 hasAuthorship W3048895265A5023187165 @default.
- W3048895265 hasAuthorship W3048895265A5064924455 @default.
- W3048895265 hasConcept C108583219 @default.
- W3048895265 hasConcept C11413529 @default.
- W3048895265 hasConcept C134306372 @default.
- W3048895265 hasConcept C149635348 @default.
- W3048895265 hasConcept C153180895 @default.
- W3048895265 hasConcept C154945302 @default.
- W3048895265 hasConcept C2776151529 @default.
- W3048895265 hasConcept C2779990667 @default.
- W3048895265 hasConcept C2781198647 @default.
- W3048895265 hasConcept C33923547 @default.
- W3048895265 hasConcept C41008148 @default.
- W3048895265 hasConcept C48372109 @default.
- W3048895265 hasConcept C50644808 @default.
- W3048895265 hasConcept C61445026 @default.
- W3048895265 hasConcept C81363708 @default.
- W3048895265 hasConcept C94375191 @default.
- W3048895265 hasConceptScore W3048895265C108583219 @default.
- W3048895265 hasConceptScore W3048895265C11413529 @default.
- W3048895265 hasConceptScore W3048895265C134306372 @default.
- W3048895265 hasConceptScore W3048895265C149635348 @default.
- W3048895265 hasConceptScore W3048895265C153180895 @default.
- W3048895265 hasConceptScore W3048895265C154945302 @default.
- W3048895265 hasConceptScore W3048895265C2776151529 @default.
- W3048895265 hasConceptScore W3048895265C2779990667 @default.
- W3048895265 hasConceptScore W3048895265C2781198647 @default.
- W3048895265 hasConceptScore W3048895265C33923547 @default.
- W3048895265 hasConceptScore W3048895265C41008148 @default.
- W3048895265 hasConceptScore W3048895265C48372109 @default.
- W3048895265 hasConceptScore W3048895265C50644808 @default.
- W3048895265 hasConceptScore W3048895265C61445026 @default.
- W3048895265 hasConceptScore W3048895265C81363708 @default.
- W3048895265 hasConceptScore W3048895265C94375191 @default.
- W3048895265 hasIssue "3" @default.
- W3048895265 hasLocation W30488952651 @default.
- W3048895265 hasOpenAccess W3048895265 @default.
- W3048895265 hasPrimaryLocation W30488952651 @default.
- W3048895265 hasRelatedWork W2731899572 @default.
- W3048895265 hasRelatedWork W2999805992 @default.
- W3048895265 hasRelatedWork W3011074480 @default.
- W3048895265 hasRelatedWork W3116150086 @default.
- W3048895265 hasRelatedWork W3133861977 @default.
- W3048895265 hasRelatedWork W4200173597 @default.
- W3048895265 hasRelatedWork W4291897433 @default.
- W3048895265 hasRelatedWork W4311401716 @default.
- W3048895265 hasRelatedWork W4312417841 @default.
- W3048895265 hasRelatedWork W4321369474 @default.
- W3048895265 hasVolume "10" @default.
- W3048895265 isParatext "false" @default.
- W3048895265 isRetracted "false" @default.
- W3048895265 magId "3048895265" @default.
- W3048895265 workType "article" @default.