[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-327556-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-327556-en":131},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","federated-learning-for-object-detection-enabling-collaborative-drone-learning-without-centralizing-data","Federated Learning for Object Detection - Enabling Collaborative Drone Learning Without Centralizing Data","","Object detection underpins AI perception for safety-critical drone and edge-vision settings, yet strong detectors usually require large, continuously updated datasets that are impractical to centralize due to privacy, regulation, storage, and bandwidth constraints. This work applies Federated Learning for collaborative drone learning while keeping image data local. A federated object detection pipeline is implemented on the KIIT-MiTA dataset using the Sherpa.ai FL platform and compared with single-drone and centralized baselines. Results keep performance close to centralized training and far above single-drone training, with the YOLO nano model delivering relative mAP gains of 52.89% at IoU 0.50 and 67.80% across 0.50:0.95.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/federated-learning-for-object-detection-enabling-collaborative-drone-learning-without-centralizing-data/327556/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/federated-learning-for-object-detection-enabling-collaborative-drone-learning-without-centralizing-data/327556.png","ImageObject",300,407,{"name":42,"@type":43},"Maeve","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-24","2026-09-21",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",15,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"Why is centralized training difficult for drone object detection?","Question",{"text":63,"@type":64},"Centralized training typically requires aggregating large aerial image datasets, but privacy regulations, proprietary concerns, limited connectivity, and storage or bandwidth limits make pooling data infeasible in many drone deployments.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How does the proposed approach enable collaborative learning without data centralization?",{"text":68,"@type":64},"It applies Federated Learning so drones collaboratively train a global object detection model while keeping local image data private and local rather than exchanging raw data.",{"name":70,"@type":61,"acceptedAnswer":71},"How does federated learning performance compare with baselines?",{"text":72,"@type":64},"Experiments show the federated approach remains close to centralized training while substantially improving over single-drone training; the best lightweight model (YOLO26 nano) achieves relative gains of 52.89% (mAP@0.50) and 67.80% (mAP@0.50:0.95).","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},327556,1790270100,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,127],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":25,"category_name":129,"show_sort_weight":97,"slug":130},19,"General","general",{"code":4,"msg":82,"data":132},{"doc_id":79,"user_id":133,"nickname":42,"user_avatar":134,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":118,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":144,"read_time":104,"preview_page_count":4},5909877438554,"https://ap-avatar.wpscdn.com/avatar/5600025385ad2bf12a7?_k=1778553567797529272","Federated Learning for Object Detection: Enabling Collaborative Drone Learning Without Centralizing Data  \n[Sherpa.ai](Sherpa.ai)  \narXiv :2607 .02636v 1 [ cs .LG] 2 Jul 2026  \n[research@sherpa.ai](research@sherpa.ai)  \nAbstract  \nObject detection is a fundamental capability for AI-driven perception in safety-critical drone and edge-vision systems, including disaster response, operational security environments, infrastructure monitoring, and defense applications. Robust model performance in such environments depends on large, continuously updated datasets. However, training highperforming detectors typically requires centralizing aerial imagery, which raises privacy, regulatory, storage, and bandwidth challenges. This is especially problematic in distributed drone deployments, where visual data is generated onboard and is often impractical or undesirable to transfer to a centralized infrastructure.  \nIn this work, we apply Federated Learning (FL) for object detection, enabling drones to improve a shared model while keeping image data local and private. We implement a federated object detection pipeline using the [Sherpa.ai](Sherpa.ai) FL platform on the KIIT-MiTA dataset, and compare it with Single-drone and Centralized baselines using mean Average Precision (mAP) at IoU thresholds of 0.50 and 0.50-0.95 . In our experiments, the proposed FL approach remains close to Centralized training while dramatically improving over Singledrone training, with the best lightweight model (YOLO26 nano), suitable for deployment even on very limited edge infrastructure, achieving relative gains of 52 .89% and 67.80% in mAP@0.50 and mAP@0.50:0.95, respectively. These results show that FL enables scalable, high-performing, and privacy-preserving object detection across distributed drone fleets without data centralization.  \nFigure 1: Global map illustrating drone mAP@0.50 for the best federated YOLO model.  \nFL for Object Detection: Enabling Collaborative Drone Learning Without Centralizing Data  \n1 Introduction  \nObject detection is a fundamental problem in computer vision that involves identifying and localizing instances of semantic objects of a certain class within an image or video. It serves as a crucial component in a wide range of applications such as drone-based surveillance, search and rescue, disaster response, infrastructure inspection, precision agriculture, and robotics. Recent advancements in deep learning have significantly improved the performance of object detection systems. Models such as Faster R-CNN [1], YOLO [2, 3 , 4], and RetinaNet [5] have become state-of-the-art benchmarks, achieving remarkable accuracy and efficiency in various settings.  \nIn drone vision, object detection is particularly important because unmanned aerial vehicles must operate in dynamic and heterogeneous environments while relying on accurate real-time perception. Compared with ground-based imagery, aerial imagery introduces additional challenges, including small object sizes, viewpoint variations, scale changes due to altitude, cluttered backgrounds, camera motion, and occlusions [6] . These factors make robust detection from drone platforms especially demanding and increase the importance of training with large, diverse, and well-annotated datasets.  \nDespite recent progress, several challenges persist in object detection. These include issues related to class imbalance, occlusion, variations in scale and aspect ratio of objects, as well as the need for large, diverse, and high-quality annotated datasets for model training [7] . Furthermore, deploying robust object detection systems in real-world drone scenarios often requires adapting models to heterogeneous data distributions, which can vary significantly across geographic regions, weather conditions, sensors, flight altitudes, and mission profiles. Addressing domain shift and ensuring model generalization under such variability remain key open problems [8] .  \nOne of the critical obstacles to furt","cbCaiefJ6mUWEnGe","https://ap.wps.com/l/cbCaiefJ6mUWEnGe","pdf",5020884,"English","# Abstract\n# 1 Introduction\n## 1.1 Motivation","[{\"question\":\"Why is centralized training difficult for drone object detection?\",\"answer\":\"Centralized training typically requires aggregating large aerial image datasets, but privacy regulations, proprietary concerns, limited connectivity, and storage or bandwidth limits make pooling data infeasible in many drone deployments.\"},{\"question\":\"How does the proposed approach enable collaborative learning without data centralization?\",\"answer\":\"It applies Federated Learning so drones collaboratively train a global object detection model while keeping local image data private and local rather than exchanging raw data.\"},{\"question\":\"How does federated learning performance compare with baselines?\",\"answer\":\"Experiments show the federated approach remains close to centralized training while substantially improving over single-drone training; the best lightweight model (YOLO26 nano) achieves relative gains of 52.89% (mAP@0.50) and 67.80% (mAP@0.50:0.95).\"}]","Federated Learning for Object Detection - Enabling Collaborative Drone Learning Without Centralizing Data | PDF",1789977526]