Matches in SemOpenAlex for { <https://semopenalex.org/work/W3207877247> ?p ?o ?g. }
- W3207877247 endingPage "10820" @default.
- W3207877247 startingPage "10808" @default.
- W3207877247 abstract "Most deep-learning-based target detection methods have high computational complexity and memory consumption, and they are difficult to deploy on edge devices with limited computing resources and memory. To tackle this problem, this article proposes to learn a lightweight detector named Light-YOLOv4, and it is obtained from YOLOv4 through model compression. To this end, first, we perform sparsity training by applying L1 regularization to the channel scaling factors, so the less important channels and layers can be found. Second, channel pruning and layer pruning are enforced on the network to prune the less important parts, which could significantly reduce network's width and depth. Third, the pruned model is retrained with a knowledge distillation method to improve the detection accuracy. Fourth, the model is quantized from FP32 to FP16, and it could further accelerate the model with almost no loss of detection accuracy. Finally, to evaluate Light-YOLOv4's performance on edge devices, Light-YOLOv4 is deployed on NVIDIA Jetson TX2. Experiments on the SAR ship detection dataset (SSDD) demonstrate that the model size, parameter size, and FLOPs of Light-YOLOv4 have been reduced by 98.63%, 98.66%, and 91.30% compared with YOLOv4, and the detection speed has been increased to 4.2×. While the detection accuracy of Light-YOLOv4 is only slightly reduced, for example, the mAP has only reduced by 0.013. Besides, experiments on the Gaofen Airplane dataset also prove the feasibility of Light-YOLOv4. Moreover, compared with other state-of-the-art methods, such as SSD and FPN, Light-YOLOv4 is more suitable for edge devices." @default.
- W3207877247 created "2021-10-25" @default.
- W3207877247 creator A5031620469 @default.
- W3207877247 creator A5033730774 @default.
- W3207877247 creator A5062810392 @default.
- W3207877247 creator A5072325419 @default.
- W3207877247 creator A5082887216 @default.
- W3207877247 creator A5089424445 @default.
- W3207877247 date "2021-01-01" @default.
- W3207877247 modified "2023-10-14" @default.
- W3207877247 title "Light-YOLOv4: An Edge-Device Oriented Target Detection Method for Remote Sensing Images" @default.
- W3207877247 cites W2233116163 @default.
- W3207877247 cites W2531409750 @default.
- W3207877247 cites W2565639579 @default.
- W3207877247 cites W2570343428 @default.
- W3207877247 cites W2782389627 @default.
- W3207877247 cites W2803867573 @default.
- W3207877247 cites W2884367402 @default.
- W3207877247 cites W2887463361 @default.
- W3207877247 cites W2896988829 @default.
- W3207877247 cites W2921271497 @default.
- W3207877247 cites W2931948385 @default.
- W3207877247 cites W2940726923 @default.
- W3207877247 cites W2955454611 @default.
- W3207877247 cites W2957667237 @default.
- W3207877247 cites W2961699889 @default.
- W3207877247 cites W2961882198 @default.
- W3207877247 cites W2962851801 @default.
- W3207877247 cites W2963037989 @default.
- W3207877247 cites W2963064196 @default.
- W3207877247 cites W2963346150 @default.
- W3207877247 cites W2963351448 @default.
- W3207877247 cites W2963857746 @default.
- W3207877247 cites W2969666521 @default.
- W3207877247 cites W2982605490 @default.
- W3207877247 cites W2988603195 @default.
- W3207877247 cites W3006898026 @default.
- W3207877247 cites W3006904694 @default.
- W3207877247 cites W3012561096 @default.
- W3207877247 cites W3028151507 @default.
- W3207877247 cites W3032483924 @default.
- W3207877247 cites W3032837604 @default.
- W3207877247 cites W3033996275 @default.
- W3207877247 cites W3042011474 @default.
- W3207877247 cites W3082766779 @default.
- W3207877247 cites W3107474049 @default.
- W3207877247 cites W3128476715 @default.
- W3207877247 cites W3128804268 @default.
- W3207877247 cites W3160457128 @default.
- W3207877247 cites W3164012963 @default.
- W3207877247 cites W3169512402 @default.
- W3207877247 cites W3184840388 @default.
- W3207877247 cites W639708223 @default.
- W3207877247 doi "https://doi.org/10.1109/jstars.2021.3120009" @default.
- W3207877247 hasPublicationYear "2021" @default.
- W3207877247 type Work @default.
- W3207877247 sameAs 3207877247 @default.
- W3207877247 citedByCount "25" @default.
- W3207877247 countsByYear W32078772472022 @default.
- W3207877247 countsByYear W32078772472023 @default.
- W3207877247 crossrefType "journal-article" @default.
- W3207877247 hasAuthorship W3207877247A5031620469 @default.
- W3207877247 hasAuthorship W3207877247A5033730774 @default.
- W3207877247 hasAuthorship W3207877247A5062810392 @default.
- W3207877247 hasAuthorship W3207877247A5072325419 @default.
- W3207877247 hasAuthorship W3207877247A5082887216 @default.
- W3207877247 hasAuthorship W3207877247A5089424445 @default.
- W3207877247 hasBestOaLocation W32078772471 @default.
- W3207877247 hasConcept C115961682 @default.
- W3207877247 hasConcept C127313418 @default.
- W3207877247 hasConcept C154945302 @default.
- W3207877247 hasConcept C162307627 @default.
- W3207877247 hasConcept C193536780 @default.
- W3207877247 hasConcept C31972630 @default.
- W3207877247 hasConcept C41008148 @default.
- W3207877247 hasConcept C62649853 @default.
- W3207877247 hasConcept C9417928 @default.
- W3207877247 hasConceptScore W3207877247C115961682 @default.
- W3207877247 hasConceptScore W3207877247C127313418 @default.
- W3207877247 hasConceptScore W3207877247C154945302 @default.
- W3207877247 hasConceptScore W3207877247C162307627 @default.
- W3207877247 hasConceptScore W3207877247C193536780 @default.
- W3207877247 hasConceptScore W3207877247C31972630 @default.
- W3207877247 hasConceptScore W3207877247C41008148 @default.
- W3207877247 hasConceptScore W3207877247C62649853 @default.
- W3207877247 hasConceptScore W3207877247C9417928 @default.
- W3207877247 hasFunder F4320321001 @default.
- W3207877247 hasLocation W32078772471 @default.
- W3207877247 hasLocation W32078772472 @default.
- W3207877247 hasOpenAccess W3207877247 @default.
- W3207877247 hasPrimaryLocation W32078772471 @default.
- W3207877247 hasRelatedWork W2045615376 @default.
- W3207877247 hasRelatedWork W2048756857 @default.
- W3207877247 hasRelatedWork W2164309059 @default.
- W3207877247 hasRelatedWork W2284111848 @default.
- W3207877247 hasRelatedWork W2347743143 @default.
- W3207877247 hasRelatedWork W2348077077 @default.
- W3207877247 hasRelatedWork W2412051338 @default.