Recent Research Progress on Ground-to-Air Vision-Based Anti-UAV Detection and Tracking Methodologies: A Review

Drones, Journal Year: 2025, Volume and Issue: 9(1), P. 58 - 58

Published: Jan. 15, 2025

Unmanned Aerial Vehicles (UAVs) are increasingly gaining popularity, and their consistent prevalence in various applications such as surveillance, search rescue, environmental monitoring requires the development of specialized policies for UAV traffic management. Integrating this novel aerial into existing airspace frameworks presents unique challenges, particularly regarding safety security. Consequently, there is an urgent need robust contingency management systems, Anti-UAV technologies, to ensure safe air traffic. This survey paper critically examines recent advancements ground-to-air vision-based detection tracking methodologies, addressing many challenges inherent tracking. Our study algorithms, outlining operational principles, advantages, disadvantages. Publicly available datasets specifically designed research also thoroughly reviewed, providing insights characteristics suitability. Furthermore, explores systems being developed deployed globally, evaluating effectiveness facilitating integration small UAVs low-altitude airspace. The aims provide researchers with a well-rounded understanding field by synthesizing current trends, identifying key technological gaps, highlighting promising directions future technologies.

Language: Английский

---

HSP-YOLOv8: UAV Aerial Photography Small Target Detection Algorithm

Drones, Journal Year: 2024, Volume and Issue: 8(9), P. 453 - 453

Published: Sept. 2, 2024

To address the larger numbers of small objects and issues occlusion clustering in UAV aerial photography, which can lead to false positives missed detections, we propose an improved object detection algorithm for scenarios called YOLOv8 with tiny prediction head Space-to-Depth Convolution (HSP-YOLOv8). Firstly, a specifically targets is added provide higher-resolution feature mapping, enabling better predictions. Secondly, designed (SPD-Conv) module mitigate loss target information enhance robustness information. Lastly, soft non-maximum suppression (Soft-NMS) used post-processing stage improve accuracy by significantly reducing results. In experiments on Visdrone2019 dataset, increased precision mAP0.5 mAP0.5:0.95 values 11% 9.8%, respectively, compared baseline model YOLOv8s.

Language: Английский

---

A Lightweight Drone Detection Method Integrated into a Linear Attention Mechanism Based on Improved YOLOv11

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(4), P. 705 - 705

Published: Feb. 19, 2025

The timely and accurate detection of unidentified drones is vital for public safety. However, the unique characteristics in complex environments varied postures they may adopt during approach present significant challenges. Additionally, deep learning algorithms often require large models substantial computational resources, limiting their use on low-capacity platforms. To address these challenges, we propose LAMS-YOLO, a lightweight drone method based linear attention mechanisms adaptive downsampling. model’s design, inspired by CPU optimization, reduces parameters using depthwise separable convolutions efficient activation functions. A novel mechanism, incorporating an LSTM-like gating system, enhances semantic extraction efficiency, improving performance scenarios. Building insights from dynamic convolution multi-scale fusion, new downsampling module developed. This efficiently compresses features while retaining critical information. improved bounding box loss function introduced to enhance localization accuracy. Experimental results demonstrate that LAMS-YOLO outperforms YOLOv11n, achieving 3.89% increase mAP 9.35% reduction parameters. model also exhibits strong cross-dataset generalization, striking balance between accuracy efficiency. These advancements provide robust technical support real-time monitoring.

Language: Английский

---

High-Precision Localization Tracking and Motion State Estimation of Ground-Based Moving Target Utilizing Unmanned Aerial Vehicle High-Altitude Reconnaissance

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(5), P. 735 - 735

Published: Feb. 20, 2025

This paper focuses on the problem of ground-motion target localization tracking and motion state estimation for high-altitude reconnaissance using fixed-wing UAVs. Our goal is to accurately locate track ground-moving targets estimate their visible light images, laser measurements distance, UAV position attitude information. Firstly, this uses detection model YOLOv8 obtain pixel positions, combined with measurement data, establish geolocalization target. Secondly, a algorithm hierarchical filtering proposed, performs optoelectronic loads separately. Using range sensor as constraints, load angle quantities are involved together in estimating ground state, resulting improved accuracy estimation. The experimental data show that reduces error by at least 7.5 m 0.8 m/s compared other algorithms.

Language: Английский

---

FQDNet: A Fusion-Enhanced Quad-Head Network for RGB-Infrared Object Detection

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(6), P. 1095 - 1095

Published: March 20, 2025

RGB-IR object detection provides a promising solution for complex scenarios, such as remote sensing and low-light environments, by leveraging the complementary strengths of visible infrared modalities. Despite significant advancements, two key challenges remain: (1) effectively integrating multi-modal features within lightweight frameworks to enable real-time performance (2) fully utilizing multi-scale features, which are crucial detecting objects varying sizes but often underexploited, leading suboptimal accuracy. To address these challenges, we propose FQDNet, novel network that integrates an optimized fusion strategy with Quad-Head framework. enhance feature fusion, introduce Channel Swap SCDown Block (CSSB) initial interaction Spatial Attention Fusion Module (SCAFM) further refine integration features. improve utilization, designed Dynamic-Weight-based Detector (DWQH), dynamically assigns weights different scales, enabling adaptive enhancing representation. This mechanism significantly improves performance, particularly small objects. Furthermore, ensure applicability, incorporate optimizations, including Partial Cross-Stage Pyramid (PCSP) modules, reduce computational complexity while maintaining high FQDNet was evaluated on three public datasets—M3FD, VEDAI, LLVIP—achieving mAP@[0.5:0.95] gains 4.4%, 3.5%, 3.1% over baseline, only 0.4 M increase in parameters 5.5 GFLOPs overhead. Compared state-of-the-art algorithms, our method strikes better balance between accuracy efficiency exhibiting strong robustness across diverse scenarios.

Language: Английский

---

Efficient and lightweight deep learning model for enhanced ship detection in maritime surveillance

Ocean Engineering, Journal Year: 2025, Volume and Issue: 328, P. 121085 - 121085

Published: April 4, 2025

Language: Английский

---

Improved YOLO for long range detection of small drones

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 10, 2025

Language: Английский

---

Design of UAV target detection network based on deep feature fusion and optimization with small targets in complex contexts

Neurocomputing, Journal Year: 2025, Volume and Issue: unknown, P. 130207 - 130207

Published: April 1, 2025

Language: Английский

---

XAI-Based Framework for Protocol Anomaly Classification and Identification to 6G NTNs with Drones

Drones, Journal Year: 2025, Volume and Issue: 9(5), P. 324 - 324

Published: April 23, 2025

Although deep learning (DL) methods are effective for detecting protocol attacks involving drones in sixth-generation (6G) nonterrestrial networks (NTNs), classifying novel and identifying anomalous sequences remain challenging. The internal capture processes matching results of DL models useful addressing these issues. key challenges involve obtaining this information from DL-based anomaly detection methods, using to establish new classifications uncovered tracing the input back sequences. Therefore, paper, we propose an interpretable classification identification method 6G NTN protocols. We design framework In particular, introduce explainable artificial intelligence (XAI) techniques obtain information, including process, a collaborative approach different utilize information. also self-evolving proposed classify attacks. rule baseline approaches made transparent work synergistically extract learn fingerprint features Furthermore, online identify sequences; intrinsic is based on two-layer neural network (DNN) model. simulation show that can be effectively used sequences, with precision increasing by maximum 32.8% at least 26%, respectively, compared existing methods.

Language: Английский

---

Simplified LSL-Net Architecture for Unmanned Aerial Vehicle Detection in Real-Time

Technologies, Journal Year: 2025, Volume and Issue: 13(5), P. 177 - 177

Published: May 1, 2025

Given the growth of unmanned aerial vehicles (UAVs), their detection has become a recent and complex problem. The literature addressed this problem by applying traditional computer vision algorithms and, more recently, deep learning architectures, which, while proven effective than previous ones, are computationally expensive. In paper, following approach we propose simplified LSL-Net-based architecture for UAV detection. This integrates ability to track detect UAVs using convolutional neural networks. biggest challenge lies in creating model that allows us obtain good results without requiring considerable computational resources. To address problem, built on successful LSL-Net architecture. We introduce dilated convolutions achieve lower-cost with capabilities. Experiments demonstrate our performs well limited resources, reaching 98% accuracy detecting UAVs.

Language: Английский

---