Yazar "Bayraktar, Irem" seçeneğine göre listele
Listeleniyor 1 - 16 / 16
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe Assessing YOLOv8 Performance and Limitations in Aircraft and Ship Detection from Satellite Images(Institute of Electrical and Electronics Engineers Inc., 2024) Etnubarhi, Nahom; Pirehen, Hale Filiz; Nagpal, Sujal; Bayraktar, IremAccurate detection of aircraft and ships from remote sensing images is crucial for various applications such as maritime surveillance, border security, and disaster management. In this study, we present a comprehensive evaluation of the performance of the YOLOv8 model in detecting aircraft and ships under diverse environmental conditions. Through extensive testing and analysis, we highlight the algorithm's capabilities and limitations in discerning objects amidst challenging backgrounds, varying lighting conditions, and complex contextual factors. Our results reveal notable successes in both aircraft and ship detection, alongside inherent challenges such as color similarity, shadow formations, and environmental turbulence. Additionally, performance metrics illustrate the model's superior efficacy in ship detection compared to aircraft detection, indicating the algorithm's adaptability to different object categories. This study contributes key insights into the capabilities of object detection algorithms in remote sensing applications, paving the way for further advancements in the field of aerial and maritime surveillance. © 2024 IEEE.Öğe Assessment of YOLO11 for Ship Detection in SAR Imagery under Open Ocean and Coastal Challenges(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Bayraktar, IremShip detection from Synthetic Aperture Radar (SAR) images plays a crucial role in maritime surveillance and safety. This study focuses on evaluating the performance of the latest state-of-The-Art YOLO algorithm, YOLO11, for ship detection, particularly because it has not been tested on SAR images prior to this research. YOLO11 was selected for its recent release and potential improvements over previous iterations. To assess its effectiveness, the algorithm is compared with earlier YOLO versions using SAR imagery. The dataset is categorized into two subsets: open ocean images and coastal images, where distinguishing ships from coastal structures presents a significant challenge. The advantages and limitations of YOLO11 are thoroughly examined through a comparative analysis with its predecessors. Results indicate that YOLO11 outperforms earlier versions in most scenarios, particularly excelling in open ocean environments. Although ship detection from SAR images is inherently difficult, YOLO11 achieves promising Precision, Recall, and mAP values of 0.865, 0.813, and 0.792, respectively. Its performance in open ocean images exceeds these average values, highlighting YOLO11's efficacy in maritime surveillance. However, performance in coastal images is lower, with YOLOv10 performing closely to YOLO11 in these cases. The findings underscore YOLO11's effectiveness for ship detection from SAR images, showcasing its enhanced ability to detect ships in challenging environments and emphasizing its relevance for future maritime applications. © 2024 IEEE.Öğe Boosting Aircraft Monitoring and Security through Ground Surveillance Optimization with YOLOv9(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Bayraktar, IremThe integration of object detection algorithms into aircraft tracking and ground surveillance systems presents a myriad of security advantages, bolstering the protection of critical infrastructure. These algorithms are instrumental in enforcing access control measures by continuously monitoring and discerning between authorized and unauthorized access to parked aircraft. With ongoing refinements, they serve as crucial components of intrusion detection systems, promptly alerting security personnel to any suspicious or unauthorized activities in the vicinity of grounded aircraft. Effective training of detection algorithms enhances their analytical capabilities, enabling them to discern between routine operations and security-threatening situations with greater precision. Notably, one of the pivotal applications lies in supporting digital forensic investigations, as these algorithms provide detailed activity logs, facilitating comprehensive post-incident analyses and bolstering forensic efforts to understand security incidents or breaches. In this investigation, we assessed the efficacy of the YOLOv9 detection algorithm in identifying aircraft situated on the ground surface. Furthermore, we highlight the significance of satellite imagery in dataset acquisition for object detection algorithms, particularly emphasizing the role of Low-Earth-Orbit (LEO) satellites in real-time image acquisition. Through this comprehensive analysis, we underscore the pivotal role of YOLOv9 in enhancing security measures and compliance with aviation security standards and regulations, ultimately fortifying the security posture within aviation environments. © 2024 IEEE.Öğe Challenges and Advances in UAV-Based Vehicle Detection Using YOLOv9 and YOLOv10(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Dmytrovych, Petro; Bayraktar, Irem; Anatoliyovych, OlehAerial imaging and object detection with unmanned aerial vehicle (UAV) systems present unique challenges, including varying altitudes, dynamic backgrounds, and changes in lighting and weather conditions. These factors complicate the detection process, demanding robust and adaptive algorithms. Furthermore, the need for real-time processing in UAV applications imposes stringent requirements on computational efficiency and resource management. This study presents a comparative analysis of two cutting-edge object detection algorithms, YOLOv9 and YOLOv10, specifically tailored for vehicle detection in UAVcaptured traffic images. Leveraging a custom dataset derived from UAV aerial imaging, both algorithms were trained and evaluated to assess their performance in terms of speed and accuracy. The experimental results reveal that while YOLOv9 demonstrates a marginally superior inference speed, YOLOv10 excels slightly in detection accuracy. © 2024 IEEE.Öğe Comparative Performance Analysis of YOLOv5 and YOLOv8 for Aerial Surveillance in Smart Traffic Management(Institute of Electrical and Electronics Engineers Inc., 2024) Nagpal, Sujal; Etnubarhi, Nahom; Pirehen, Hale Filiz; Bayraktar, IremDetection and classification of vehicles in various traffic conditions are critical to advances in traffic monitoring and autonomous vehicle technologies. This study addresses the need for more accurate and efficient vehicle detection algorithms by evaluating the performance of YOLOv5 and YOLOv8 object detection models. The smallest and largest submodels (YOLOv5s, YOLOv5x, YOLOv8s, and YOLOv8x) were tested using images captured under extreme conditions, including nighttime and high-altitude drone images. The methodology involves systematically comparing the detection accuracy and inference speed of these models. The results revealed that the 'x' submodels outperformed the 's' submodels in terms of detection accuracy across various scenarios. In particular, YOLOv8x demonstrated high detection and classification performance in complex conditions, such as images with varying light filters and dark backgrounds. Conversely, the 's' submodels, although significantly faster, exhibited lower detection accuracy. Additionally, YOLOv8 models outperformed YOLOv5 models across all performance metrics, with YOLOv8x providing the highest accuracy. The study concludes that while there is a trade-off between detection accuracy and inference speed, the YOLOv8x model offers the best balance for applications that require high precision in vehicle detection. These findings provide valuable insights into selecting the appropriate model according to specific application requirements, ultimately contributing to the development of traffic monitoring and autonomous vehicle systems. © 2024 IEEE.Öğe Comparative Performance of YOLOv9 and YOLOv10 for Vehicle Detection Towards Real-Time Traffic Surveillance with UAVs(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Bayraktar, IremIntelligent transportation systems (ITS) have gained significant traction since the 1980s and 1990s, driven by technological advancements and increasing urbanization, which have caused intense transportation challenges. Drone systems, with their superior imaging capabilities, offer critical solutions for swift traffic surveillance, surpassing traditional monitoring systems. In this context, object recognition is crucial, and the YOLO algorithm stands out for its speed and efficiency. This study conducts a detailed performance evaluation of the YOLOv9 and YOLOv10 networks for motor vehicle classification through aerial images captured by drone platforms. Datasets were created from these UAV-based traffic images, and the performances of both algorithms were measured and compared. The results were analyzed to highlight YOLOv9 and YOLOv10's strengths and drawbacks. Additionally, the study discusses qualitative aspects, including advantages, disadvantages, and potential improvements for both algorithms in aerial traffic monitoring. © 2024 IEEE.Öğe Harnessing UAV Technology and YOLOv9 Algorithm for Real-Time Forest Fire Detection(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Bayraktar, IremClimate change poses a pressing global challenge, precipitating significant natural disasters such as uncontrollable forest fires, droughts, and floods. These calamities exact severe tolls on human lives, ecosystems, and economies, prompting nations to seek proactive strategies for early forest fire intervention. While natural elements like dried vegetation and pine cones serve as primary fire catalysts, human activities exacerbate the incidence of forest fires, necessitating robust detection methods. Traditional approaches, including watchtowers and optical smoke detection systems, offer partial solutions, but real-time detection remains elusive. Unmanned Aerial Vehicles (UAVs) equipped with advanced sensors emerge as game-changers in forest fire detection, offering rapid and accurate surveillance over expansive forested areas. UAVs equipped with thermal imaging cameras detect fire heat signatures amidst dense foliage, aiding timely response efforts. Furthermore, their agility and maneuverability enable access to hazardous terrains, enhancing situational awareness and post- fire assessment. This study explores the efficacy of the YOLOv9 algorithm for real-time fire detection using UAV-captured imagery, showcasing its potential to revolutionize forest fire management. Leveraging advancements in computer vision technology, the YOLOv9 algorithm promises accelerated and precise fire detection, underscoring its pivotal role in mitigating the devastating impact of forest fires on ecosystems and communities. © 2024 IEEE.Öğe Improving Coastal and Port Management in Smart Cities with UAVs and Deep Learning(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Bayraktar, IremEfficient coast and harbor management is integral to the vitality, sustainability, and resilience of smart cities. With bustling harbors serving as vital hubs of commerce, trade, and tourism, optimizing port operations is paramount for economic growth and prosperity. Smart technologies play a pivotal role in this optimization, leveraging advanced sensor networks, real-time monitoring systems, and predictive analytics to enhance safety, mitigate environmental risks, and improve overall efficiency. Additionally, smart coastal management strategies focus on preserving ecosystems, mitigating climate change impacts, and safeguarding against natural disasters. Aerial imagery, facilitated by Unmanned Aerial Vehicles (UAVs) equipped with high-resolution cameras and sensors, provides comprehensive insights into coastal dynamics, harbor operations, and environmental conditions. These images enable efficient monitoring of coastal areas, ports, and harbors, capturing crucial information for informed decision-making in coastal management and port operations. Object detection, particularly in ship detection, stands as a transformative technology for enhancing coastal and harbor management within smart cities. Leveraging advanced algorithms and high-resolution aerial imagery, ship detection systems offer real-time monitoring crucial for optimizing maritime operations and ensuring port security. Object detection algorithms, particularly Faster R-CNN, have shown promise in accurately detecting ships in aerial imagery, offering valuable insights for harbor planning and infrastructure development. This study focuses on utilizing the Faster R-CNN detection algorithm for ship detection in coastal and harbor environments, highlighting its potential to bolster security applications and contribute to the resilience of smart city infrastructure. Through rigorous evaluation and optimization, this research aims to enhance the effectiveness of ship detection systems in safeguarding coastal and harbor environments within smart cities. ©2024 IEEE.Öğe Integrating UAV-Based Aerial Monitoring and SSD for Enhanced Traffic Management in Smart Cities(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Bayraktar, IremIn the context of smart cities, transportation stands out as a critical aspect, showcasing innovative solutions to tackle congestion, emissions, and overall mobility challenges. From intelligent traffic management systems to the proliferation of shared mobility services, these cities prioritize efficiency and sustainability. Moreover, they foster the development of electric and autonomous vehicles, enhancing safety and reducing carbon footprints. Intelligent transportation systems are crucial in addressing transportation challenges within smart cities. These systems harness cutting-edge technology to optimize efficiency, safety, and sustainability by enabling dynamic traffic management and promoting multimodal connectivity. Object detection technology further enhances traffic management by providing real-time data for proactive intervention and optimization of traffic patterns. Object detection algorithms revolutionize traffic management by accurately identifying and tracking vehicles in real-time. Additionally, unmanned aerial vehicles (UAVs) serve as powerful tools for aerial traffic monitoring, providing valuable real-time imagery for object detection algorithms. This study focuses on enhancing ITS within smart cities through the integration of UAV-based aerial monitoring and the SSD detection algorithm for vehicle detection. By compiling a large dataset using vehicle images acquired through UAV aerial monitoring, the study demonstrates robust and real-time vehicle detection capabilities. The integration of UAV-based aerial monitoring with SSD vehicle detection holds immense potential for improving overall transportation within cities, contributing to safer, more efficient, and sustainable urban environments. ©2024 IEEE.Öğe LEO Satellites: Enhancing Connectivity and Data Collection Across Industries(Springer Science and Business Media Deutschland GmbH, 2025) Yesilkaya, Nimet Selen; Bayraktar, IremLow Earth Orbit (LEO) satellites have become essential in various industries, transforming connectivity, data collection, communication, and navigation. Orbiting closer to Earth offers advantages such as lower latency and cost-effective deployment, making LEO satellites crucial for modern industrial infrastructure. With a wide range of applications, they play a vital role in monitoring atmospheric layers and near-space environments, from communication and observation to exploration. Their consistent delivery of high-precision data over extended periods is a key strength. This study highlights the increasing importance of satellite technology, driven by advantages like lower transmission power requirements and improved coverage, especially in polar regions. Understanding these benefits is essential for maximizing satellite technology’s potential. To harness LEO satellites effectively, a thorough understanding of orbital parameters and precise control is necessary. This research thoroughly examines orbit analysis and Earth coverage, providing insights into coverage areas and capabilities. It also explores data exchange dynamics with ground stations and their practical implications. In summary, this study offers a detailed exploration of LEO satellite technology, shedding light on its diverse contributions to various sectors. The proposed strategic ground station network not only greatly extends communication durations with LEO satellites but also enables more frequent updates to satellite commands. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.Öğe Multi-Class Vehicle Detection and Classification with YOLO11 on UAV-Captured Aerial Imagery(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Dmytrovych, Petro; Bayraktar, Irem; Anatoliyovych, OlehAerial imaging and object detection using unmanned aerial vehicle (UAV) systems pose unique challenges, including varying altitudes, dynamic backgrounds, and changes in lighting and weather conditions. These factors complicate the detection process, demanding robust and adaptive algorithms. Furthermore, the need for real-time processing in UAV applications imposes stringent requirements on computational efficiency and resource management. This study presents a comparative analysis of the cutting-edge object detection algorithm YOLO11, specifically tailored for vehicle detection in UAV-captured traffic images. Using a custom dataset derived from UAV aerial imaging, the algorithm was trained and evaluated to assess both its performance in terms of speed and accuracy, and the results were compared with YOLOv10. Experimental findings indicate that while YOLOv10 achieves slightly faster inference speeds, YOLO11 offers marginally better detection accuracy. © 2024 IEEE.Öğe Performance Evaluation of YOLOv4 for Instant Object Detection in UAVs(Ieee, 2024) Keskinoglu, Gul Bahar; Yesilkaya, Nimet Seleri; Bayraktar, Irem; Kose, ErcanUnmanned Aerial Vehicles (UAVs) have become integral in various research domains due to the advantages they provide. Current UAV systems rely on Global Navigation Satellite Systems (GNSS) for flight control and sensors for obstacle detection, yet fully autonomous decision-making remains a challenge. This study evaluates the performance of YOLOv4, a Convolutional Neural Network (CNN) image recognition algorithm, for instantaneous object detection and classification in UAV-captured aerial images. The study demonstrates the applicability of YOLOv4 in real-time object detection and classification through UAV image feeds. The proposed approach advances the understanding of deploying CNNs in UAVs, offering a cost-effective solution for real-time object detection and classification, essential for autonomous UAV operations.Öğe Refining Transportation Automation with Convolutional Neural Network-Based Vehicle Detection via UAVs(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Bayraktar, IremIn response to the surge in private vehicle usage exacerbated by the pandemic, transportation infrastructure faces heightened demands for efficiency and safety. This trend, fueled by concerns over social distancing and changes in work patterns, has intensified traffic congestion in urban centers and tourist destinations. Consequently, local governments are exploring strategies such as smart traffic management systems and alternative transportation modes to alleviate congestion and promote sustainability. However, the escalating vehicular influx has led to a surge in traffic accidents and air pollution, posing significant threats to public health and ecological balance. Continual monitoring of road traffic is crucial for preempting traffic bottlenecks, averting accidents, and promoting sustainable transportation practices. Advanced technologies, including unmanned aerial vehicles (UAVs), have emerged as powerful tools for real-time traffic monitoring. Equipped with high-resolution cameras and sensor systems, UAVs facilitate efficient data acquisition from desired regions, offering advantages over traditional aerial vehicles. While UAVs present promising solutions, real-time utilization poses challenges such as complex background scenes and detection misses. Additionally, advancements in object detection algorithms, particularly the You Only Look Once (YOLO) algorithm, have revolutionized vehicle detection. This study focuses on evaluating vehicle detection using YOLOv9, the latest iteration, and compares it with YOLOv7, its predecessor, leveraging aerial monitoring via UAVs. By elucidating the innovations introduced by YOLOv9, this study contributes to enhancing traffic monitoring capabilities and promoting effective transportation management strategies. © 2024 IEEE.Öğe The Cutting-Edge YOLO11 for Advanced Aircraft Detection in Synthetic Aperture Radar (SAR) Imagery(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Bayraktar, IremThis study presents one of the first evaluations of the YOLO11 algorithm and is the first to apply it for aircraft detection from SAR images. A dataset combining SAR Aircraft Detection Dataset (SADD) images and additional web-mined images was used to train and test the model. YOLO11 achieved significant improvements in detection accuracy, with higher precision, recall, and mAP compared to earlier YOLO iterations such as YOLOv5, YOLOv8, YOLOv9, and YOLOv10. The model exhibited a balanced performance by maintaining competitive inference times while minimizing both false positives and missed detections. These results demonstrate the potential of YOLO11 for real-time applications, particularly in UAV-based surveillance systems, where both speed and accuracy are critical. © 2024 IEEE.Öğe Transforming Aircraft Detection Through LEO Satellite Imagery and YOLOv9 for Improved Aviation Safety(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Bayraktar, IremThe utilization of object detection algorithms in conjunction with satellite imagery to detect aircraft has become a crucial focus of investigation, bearing significant implications for the management of airports, as well as for the enhancement of aviation safety and security. Traditional monitoring methods, reliant on manual observation or radar systems, are laborintensive, error-prone, and may struggle to provide comprehensive coverage, particularly in vast or remote areas. In contrast, satellite imagery offers wide-area coverage and high-resolution imagery, ideal for capturing detailed views of airport facilities and surrounding areas. By coupling satellite imagery with advanced object detection algorithms like YOLO, significant improvements in aircraft detection capabilities are achievable. These algorithms streamline monitoring processes, enhance situational awareness, and enable prompt responses to potential safety or security threats, marking a paradigm shift in aircraft monitoring practices. Object detection algorithms, especially those based on deep learning techniques like convolutional neural networks (CNNs), have revolutionized object identification in images or video streams. Their adaptability to diverse environmental conditions and high accuracy make them invaluable across numerous applications, from surveillance to autonomous vehicles. Satellite imagery, particularly from Low Earth Orbit (LEO) satellites, offers exceptional detail and clarity, with shorter revisit times enabling near-real-time monitoring of dynamic events on the ground. Leveraging these advantages, this study focuses on enhancing airport security and aircraft safety through the detection of aircraft on the ground utilizing YOLOv9, chosen for its promising capabilities. While our findings demonstrate notable advancements, it is crucial to address certain limitations in the algorithm's detection capabilities to ensure its effectiveness in real-world applications. © 2024 IEEE.Öğe YOLOv9-Enabled Vehicle Detection for Urban Security and Forensics Applications(Institute of Electrical and Electronics Engineers Inc., 2024) Bakirci, Murat; Bayraktar, IremThe integration of artificial intelligence (AI) techniques in vehicle detection holds significant promise, particularly in forensic and security domains. Leveraging object detection algorithms enables real-time monitoring of vehicles by competent authorities, aiding in continuous surveillance of roads and highways for various surveillance objectives. Additionally, it streamlines tasks such as identifying stolen vehicles, tracking suspects, and enforcing traffic regulations. Object detection technology also proves invaluable in forensic analysis, aiding criminal investigations and accident reconstructions. Furthermore, it enhances security by detecting aberrant behavior and potential threats at critical infrastructure sites. Concurrently, the remarkable advancements in unmanned aerial vehicles (UAVs) have led to their widespread application across diverse domains, including traffic monitoring and intelligent transportation systems. Equipped with high-resolution cameras, UAVs offer precise imagery for vehicle detection, facilitating swift responses to incidents. This study focuses on vehicle detection from aerial urban transportation images using YOLOv9 on a UAV platform, demonstrating the feasibility and efficacy of aerial analysis for efficient vehicle detection and timely alerts to competent authorities. © 2024 IEEE.
