{"id":{"repo_id":"vilnius","oai_identifier":"oai:vu.lt:elaba:308332332"},"canonical_url":"https://search.ndltd.org/etd/vilnius/oai:vu.lt:elaba:308332332","repository":{"repo_id":"vilnius","name":"Vilnius University","base_url":"https://epublications.vu.lt/oai"},"display":{"title":"Oro objektų atpažinimas gynybos srityje /","abstract":"This bachelor thesis addresses the challenge of real-time unmanned aerial vehicle (UAV) detection, continuous multi-object tracking, and dynamic threat level classification in optical air defense systems. The aim of the study is to design and experimentally evaluate a software subsystem capable of operating stably under varying meteorological and distance conditions. The research employs a tracking-by-detection methodology, utilizing the single-stage YOLOv8s neural network alongside ByteTrack and DeepSORT tracking algorithms. To generate experimental data, the photorealistic AirSim simulation environment was used to create a dataset of 11,081 frames, encompassing tactical flight scenarios, various distances, and visual disturbances (rain, storm, fog). The experimental results demonstrated that the YOLOv8s model achieved a precision of 90.0% and an mAP50 accuracy of 71%. A comparative analysis of the tracking algorithms revealed that the appearance-based DeepSORT algorithm is highly sensitive to background noise in low-visibility conditions, leading to the accumulation of false ghost tracks. In contrast, the spatial kinematics-based ByteTrack algorithm, utilizing two-stage bounding box association, more successfully filters out noise and maintains track integrity in the far zone. The developed heuristic threat assessment module successfully classified targets (LOW, MEDIUM, HIGH) based on the instantaneous change in bounding box spatial occupancy and expansion velocity. Tested on cloud infrastructure with an NVIDIA Tesla T4 accelerator, the integrated system demonstrated a high processing speed of 75FPS, proving its viability for practical application in real-time anti-drone surveillance systems.","abstract_html":"This bachelor thesis addresses the challenge of real-time unmanned aerial vehicle (UAV) detection, continuous multi-object tracking, and dynamic threat level classification in optical air defense systems. The aim of the study is to design and experimentally evaluate a software subsystem capable of operating stably under varying meteorological and distance conditions. The research employs a tracking-by-detection methodology, utilizing the single-stage YOLOv8s neural network alongside ByteTrack and DeepSORT tracking algorithms. To generate experimental data, the photorealistic AirSim simulation environment was used to create a dataset of 11,081 frames, encompassing tactical flight scenarios, various distances, and visual disturbances (rain, storm, fog). The experimental results demonstrated that the YOLOv8s model achieved a precision of 90.0% and an mAP50 accuracy of 71%. A comparative analysis of the tracking algorithms revealed that the appearance-based DeepSORT algorithm is highly sensitive to background noise in low-visibility conditions, leading to the accumulation of false ghost tracks. In contrast, the spatial kinematics-based ByteTrack algorithm, utilizing two-stage bounding box association, more successfully filters out noise and maintains track integrity in the far zone. The developed heuristic threat assessment module successfully classified targets (LOW, MEDIUM, HIGH) based on the instantaneous change in bounding box spatial occupancy and expansion velocity. Tested on cloud infrastructure with an NVIDIA Tesla T4 accelerator, the integrated system demonstrated a high processing speed of 75FPS, proving its viability for practical application in real-time anti-drone surveillance systems.","abstract_has_math":false,"creators":["Baronas, Liudas,"],"institution":"Institutional Repository of Vilnius University","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-19T10:12:52Z","subjects":[],"languages":["lit"],"rights":["info:eu-repo/semantics/openAccess"],"rights_urls":[],"identifier_entries":[],"person_identifier_entries":[]},"links":{"outbound_url":"https://repository.vu.lt/VU:ELABAETD308332332&prefLang=en_US","outbound_label":"Repository record","outbound_source":"dc:identifier","record_url":"https://repository.vu.lt/VU:ELABAETD308332332&prefLang=en_US","record_label":"Repository record","record_kind":"record","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":"open","embargo_until":null,"access_note":null},"source_record":{"url":"https://epublications.vu.lt/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Avu.lt%3Aelaba%3A308332332","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Baronas, Liudas,"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2026"]},{"key":"dc:publisher","label":"Institution","values":["Institutional Repository of Vilnius University"]},{"key":"dc:relation","label":"Dc Relation","values":["https://epublications.vu.lt/object/elaba:308332332/308332332.pdf"]},{"key":"dc:type","label":"Dc Type","values":["info:eu-repo/semantics/bachelorThesis"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["lit"]},{"key":"dc:rights","label":"Dc Rights","values":["info:eu-repo/semantics/openAccess"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://repository.vu.lt/VU:ELABAETD308332332&prefLang=en_US"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["This bachelor thesis addresses the challenge of real-time unmanned aerial vehicle (UAV) detection, continuous multi-object tracking, and dynamic threat level classification in optical air defense systems. The aim of the study is to design and experimentally evaluate a software subsystem capable of operating stably under varying meteorological and distance conditions. The research employs a tracking-by-detection methodology, utilizing the single-stage YOLOv8s neural network alongside ByteTrack and DeepSORT tracking algorithms. To generate experimental data, the photorealistic AirSim simulation environment was used to create a dataset of 11,081 frames, encompassing tactical flight scenarios, various distances, and visual disturbances (rain, storm, fog). The experimental results demonstrated that the YOLOv8s model achieved a precision of 90.0% and an mAP50 accuracy of 71%. A comparative analysis of the tracking algorithms revealed that the appearance-based DeepSORT algorithm is highly sensitive to background noise in low-visibility conditions, leading to the accumulation of false ghost tracks. In contrast, the spatial kinematics-based ByteTrack algorithm, utilizing two-stage bounding box association, more successfully filters out noise and maintains track integrity in the far zone. The developed heuristic threat assessment module successfully classified targets (LOW, MEDIUM, HIGH) based on the instantaneous change in bounding box spatial occupancy and expansion velocity. Tested on cloud infrastructure with an NVIDIA Tesla T4 accelerator, the integrated system demonstrated a high processing speed of 75FPS, proving its viability for practical application in real-time anti-drone surveillance systems."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Oro objektų atpažinimas gynybos srityje /","Aerial object recognition in defense."]}]}],"canonical_facts":{"dc:creator":["Baronas, Liudas,"],"dc:date":["2026"],"dc:description":["This bachelor thesis addresses the challenge of real-time unmanned aerial vehicle (UAV) detection, continuous multi-object tracking, and dynamic threat level classification in optical air defense systems. The aim of the study is to design and experimentally evaluate a software subsystem capable of operating stably under varying meteorological and distance conditions. The research employs a tracking-by-detection methodology, utilizing the single-stage YOLOv8s neural network alongside ByteTrack and DeepSORT tracking algorithms. To generate experimental data, the photorealistic AirSim simulation environment was used to create a dataset of 11,081 frames, encompassing tactical flight scenarios, various distances, and visual disturbances (rain, storm, fog). The experimental results demonstrated that the YOLOv8s model achieved a precision of 90.0% and an mAP50 accuracy of 71%. A comparative analysis of the tracking algorithms revealed that the appearance-based DeepSORT algorithm is highly sensitive to background noise in low-visibility conditions, leading to the accumulation of false ghost tracks. In contrast, the spatial kinematics-based ByteTrack algorithm, utilizing two-stage bounding box association, more successfully filters out noise and maintains track integrity in the far zone. The developed heuristic threat assessment module successfully classified targets (LOW, MEDIUM, HIGH) based on the instantaneous change in bounding box spatial occupancy and expansion velocity. Tested on cloud infrastructure with an NVIDIA Tesla T4 accelerator, the integrated system demonstrated a high processing speed of 75FPS, proving its viability for practical application in real-time anti-drone surveillance systems."],"dc:format":["application/pdf"],"dc:identifier":["https://repository.vu.lt/VU:ELABAETD308332332&prefLang=en_US"],"dc:language":["lit"],"dc:publisher":["Institutional Repository of Vilnius University"],"dc:relation":["https://epublications.vu.lt/object/elaba:308332332/308332332.pdf"],"dc:rights":["info:eu-repo/semantics/openAccess"],"dc:title":["Oro objektų atpažinimas gynybos srityje /","Aerial object recognition in defense."],"dc:type":["info:eu-repo/semantics/bachelorThesis"]},"updated_at":"2026-11-19T10:12:52Z"}