[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-31230":3,"doc-seo-31230":26},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"file_id":15,"file_url":16,"file_type":17,"file_size":18,"view_count":4,"is_deleted":4,"is_public":19,"is_downloadable":19,"audit_status":19,"page_count":11,"language":20,"language_code":21,"table_of_contents":22,"faqs":23,"seo_title":13,"seo_description":14,"update_tm":24,"read_time":25},31230,4810365810221,"Aurora","https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d",8,"Research & Report","A Lightweight Hierarchical AI Model for UAV-Enabled Edge Computing With Forest-Fire Detection Use-case","UAV-based edge computing enables early, localized analysis of drone-acquired data without overloading limited onboard computation, energy, and communication resources. For forest-fire detection, CNN inference on the drone can exhaust resources and reduce practical deployability. A lightweight hierarchical AI framework is introduced, adaptively switching between a simple machine-learning model and a deeper CNN model. A multi-objective optimization models accuracy–performance tradeoffs and produces Pareto-optimal solutions by tuning a confidence-threshold hyperparameter using TOPSIS, maintaining high detection quality while reducing computational burden. ","cbCaitvJXzSonAmY","https://ap.wps.com/l/cbCaitvJXzSonAmY","pdf",727362,1,"English","en","# Introduction\n## UAV-enabled edge computing challenges\n## Forest-fire detection use-case\n## Motivation for lightweight hierarchical AI","[{\"question\":\"Why does forest-fire detection inference on a UAV create a resource problem?\",\"answer\":\"Running CNN-based inference for forest-fire recognition can exhaust the UAV’s limited onboard computation, energy, and communication resources, making resource-aware deployment difficult.\"},{\"question\":\"How does the proposed hierarchical framework reduce computation while preserving accuracy?\",\"answer\":\"It adaptively switches between a lightweight machine-learning model and an advanced CNN model, performing deeper inference only when needed rather than continuously.\"},{\"question\":\"How is the accuracy–computation tradeoff optimized in the proposed approach?\",\"answer\":\"A multi-objective optimization problem is formulated to maximize detection efficiency while minimizing computational cost, and Pareto-optimal solutions are obtained by tuning a confidence-score threshold using TOPSIS.\"}]",1779224468,20,{"code":4,"msg":27,"data":28},"ok",{"site_id":29,"language":21,"slug":30,"title":13,"keywords":31,"description":14,"schema_data":32,"social_meta":83,"head_meta":85,"extra_data":87,"updated_unix":24},105,"a-lightweight-hierarchical-ai-model-for-uav-enabled-edge-computing-with-forest-fire-detection-use-case","",{"@graph":33,"@context":82},[34,51,65],{"@type":35,"itemListElement":36},"BreadcrumbList",[37,41,45,48],{"item":38,"name":39,"@type":40,"position":19},"https://docshare.wps.com","Home","ListItem",{"item":42,"name":43,"@type":40,"position":44},"https://docshare.wps.com/document/","Document",2,{"item":46,"name":12,"@type":40,"position":47},"https://docshare.wps.com/document/research-report/",3,{"item":49,"name":13,"@type":40,"position":50},"https://docshare.wps.com/document/a-lightweight-hierarchical-ai-model-for-uav-enabled-edge-computing-with-forest-fire-detection-use-case/31230/",4,{"url":49,"name":13,"@type":52,"author":53,"headline":13,"publisher":55,"fileFormat":58,"description":14,"dateModified":59,"datePublished":59,"encodingFormat":58,"isAccessibleForFree":60,"interactionStatistic":61},"DigitalDocument",{"name":9,"@type":54},"Person",{"url":38,"name":56,"@type":57},"DocShare","Organization","application/pdf","2026-05-19",true,{"@type":62,"interactionType":63,"userInteractionCount":4},"InteractionCounter",{"@type":64},"ViewAction",{"@type":66,"mainEntity":67},"FAQPage",[68,74,78],{"name":69,"@type":70,"acceptedAnswer":71},"Why does forest-fire detection inference on a UAV create a resource problem?","Question",{"text":72,"@type":73},"Running CNN-based inference for forest-fire recognition can exhaust the UAV’s limited onboard computation, energy, and communication resources, making resource-aware deployment difficult.","Answer",{"name":75,"@type":70,"acceptedAnswer":76},"How does the proposed hierarchical framework reduce computation while preserving accuracy?",{"text":77,"@type":73},"It adaptively switches between a lightweight machine-learning model and an advanced CNN model, performing deeper inference only when needed rather than continuously.",{"name":79,"@type":70,"acceptedAnswer":80},"How is the accuracy–computation tradeoff optimized in the proposed approach?",{"text":81,"@type":73},"A multi-objective optimization problem is formulated to maximize detection efficiency while minimizing computational cost, and Pareto-optimal solutions are obtained by tuning a confidence-score threshold using TOPSIS.","https://schema.org",{"og:url":49,"og:type":84,"og:title":13,"og:site_name":56,"og:description":14},"article",{"robots":86,"canonical":49},"index,follow",{"doc_id":7,"site_id":29}]