Search Results (1,799)

Existing single-image deraining methods often exhibit limited 2D long-range dependency modeling and underexploit frequency-domain priors. To address this, we propose SFQMamba, a dual-branch deraining network based on spatial–frequency feature fusion. The CNN branch employs a Fused Enhance Block (FEB), which integrates multi-scale spatial modeling with global frequency modulation, supported by residual coupling and channel guidance, to suppress rain streaks and recover structural details. Concurrently, the Mamba branch utilizes a Spatial-Aware Selective Fusion Block (SASFB). By incorporating a four-directional scanning mechanism and adaptive path-gating, SASFB extends 1D State Space Models into the 2D domain for content-aware feature fusion. Features from both branches are hierarchically aggregated via concatenation and pointwise convolution. Experiments on the Rain13K and Raindrop datasets show that SFQMamba provides robust restoration. Compared with TransMamba, it obtains improvements of 0.12 dB in PSNR and 0.11% in SSIM, removing dense rain streaks while preserving structural and textural details. Furthermore, on the RainVisDrone benchmark, specifically the medium-rain subset, our method improves YOLOv8s detection by 0.0737 AP, 0.1060 AP${⁠}_{50}$, and 0.0897 AP${⁠}_{75}$ over degraded inputs. These results indicate that the proposed framework benefits both low-level visual restoration and downstream object perception in UAV applications. Full article

Global Navigation Satellite Systems (GNSS)-denied environments pose significant challenges for autonomous drone navigation, requiring robust visual localization systems capable of real-time performance. Existing approaches either sacrifice accuracy for speed or fail to adapt to varying flight altitudes and orientations, limiting their practical deployment. We present Real-Time Elevation and Orientation-Aware Localization Architecture (REOLA), a visual localization system that combines similarity-driven autonomous window sizing, element-wise correlation-based orientation detection, and reinforcement learning with human feedback (RLHF) enhancement for publicly available satellite imagery. On desktop hardware (i7-10700K + RTX 3070), the REOLA achieved approximately 59 FPS performance with sub-5-m accuracy across diverse flight conditions through intelligent similarity-based matching, combined with efficient MobileNet-V3 embeddings and FAISS similarity search. For embedded deployment on NVIDIA Jetson Orin Nano, the system achieved 22.5 FPS, meeting real-time requirements for autonomous drone localization. The system autonomously selects optimal window sizes corresponding to the current elevation and determines drone orientation through element-wise correlation scoring across discrete rotation angles. Enhanced through RLHF, the REOLA achieved a $97.1\%$ success rate (sub-5-m localization) while processing frames in 17 milliseconds on desktop hardware (44.4 ms on embedded hardware), providing a substantial margin over real-time requirements. The approach demonstrates particular superiority over traditional keypoint-based methods in challenging environments with repetitive patterns such as agricultural fields, rocky mountains, dense forests, and grasslands, where conventional keypoint detection struggles. We explicitly identify featureless sand dune deserts and open-sea or coastal water flights as out of scope, since the reference satellite imagery in those regimes does not contain stable landmarks. Full article

UAV-based vehicle detection faces challenges of small targets, dense distribution, and occlusions. Built upon YOLOv10n, this paper proposes YOLOv10-IAD by integrating three modules: (1) Involution convolution in the backbone to enlarge the receptive field and enhance spatial perception for small targets; (2) ACmix (Attention and Convolution Mixed) in the neck to fuse local details with global context; (3) DyHead (Dynamic Head) that recalibrates features via scale-, space-, and task-aware attention, improving localization for occluded objects. On VisDrone2019 and UAVDT datasets, YOLOv10-IAD improves mAP50 by 3.7% (to 47.2%) and 3.5% (to 52.0%), and recall by 3.1% and 2.0%, respectively, with only a modest increase in parameters (2.9 M) and computational cost. Compared to other YOLO series, it achieves a favorable trade-off between detection accuracy and computational efficiency. These advancements make it suitable for deployment on hardware onboard UAVs for real-time road vehicle detection. Full article

The increasing activity of unmanned aerial vehicles (UAVs) has intensified the demand for reliable and computationally efficient methods for passive radio-frequency (RF) signal detection. In practical RF monitoring scenarios, the environment is often non-stationary and affected by varying noise conditions. Under such circumstances, classical energy-based detectors are sensitive to noise uncertainty, while more robust approaches, such as cyclostationary analysis, require substantially higher computational resources. This work presents a burst-aware cascade method for UAV RF signal presence detection that explicitly addresses this trade-off. The proposed framework combines fast energy-based screening with temporal burst aggregation, applying spectral correlation function (SCF) analysis selectively and only when sustained signal activity is indicated. Detection is performed on fixed-length RF signal chunks, while additional segment-level duration constraints are used to characterize sustained transmissions. The method is evaluated using the publicly available DroneRF dataset and compared against six baseline detectors, including fixed-threshold energy, wavelet-based, blind cyclostationary, two adaptive energy detector variants, and a lightweight convolutional neural network. Experimental results confirm that chunk-level detection remains difficult for all considered methods. Temporal aggregation across longer intervals yields a substantial improvement: the cascade achieves $P_d$ = 1.000 and AUC = 1.000 at the segment level, matching exhaustive cyclostationary detection while reducing per-segment processing time by a factor of 2.46. An additional result is that burst-level concatenation prior to SCF estimation provides implicit coherent integration, preserving $P_d$ = 1.000 at signal amplitude reductions of up to −20 dB where standalone detectors degrade to $P_d$ = 0.995. Overall, burst-aware cascade architectures offer a practical and interpretable approach to RF-based UAV monitoring, providing a well-grounded compromise between detection reliability and computational efficiency under realistic operating conditions. Full article

Small object detection in UAV aerial imagery poses significant challenges due to sparse pixel representation and ambiguous object boundaries. Through systematic analysis, we identify three critical degradation stages during forward propagation in deep detection networks: edge attenuation during feature extraction, semantic conflict during feature fusion, and detail loss during feature reconstruction. Existing methods address these stages in isolation or implicitly, lacking collaborative and stage-aware repair strategies. To address this issue, we propose EAGLE-DET, a novel detection framework based on sparse multi-scale attention and refined transformation. Specifically, the framework comprises three core modules: (1) the Cross-stage Multi-resolution Edge Enhancement Network (CMENet), which preserves small object edge representations via adaptive high-low frequency decomposition; (2) the Attention-guided Multi-scale Feature Fusion Network (AMFFN), which resolves cross-scale semantic conflicts through pyramidal sparse attention and multi-scale spatial decoupling; (3) the Enhanced Upsampling with Channel Bridging and Spatial Coordination module (EUCBSC), which recovers spatial detail fidelity via bidirectional channel shift mixing. Extensive experiments on three benchmark datasets—VisDrone-2019, UAVDT, and DOTA1.0—demonstrate the effectiveness of EAGLE-DET, which achieves improvements of 4.5% AP50 and 2.9% AP50:95 on VisDrone-2019 over the baseline, while maintaining inference at 71.7 FPS, achieving an optimal accuracy–efficiency trade-off. Full article

In the field of remote sensing applications, multimodal object detection has emerged as an important technique for enhancing perception robustness in UAV-based scenarios. Nevertheless, RGB–IR UAV detection remains difficult: Degraded illumination destabilizes shallow representations and weakens local discriminative cues, while spatial inconsistencies and fluctuating modality reliability further hinder cross-modal interaction. In addition, existing methods, which often depend on global illumination estimation or simplistic fusion schemes, struggle to jointly maintain contextual stability, reliable cross-modal interaction, and compact discriminative representations in complex aerial scenes. To address these issues, this paper proposes LDSDet, an RGB–IR multimodal UAV object detector for challenging illumination conditions. Specifically, LDSDet integrates three complementary modules: a Long-range Aware Residual Convolution (LARC) module that enhances contextual perception and stabilizes shallow features; a Dynamic Attention-based Cross-modal Fusion (DACF) block that performs spatially adaptive RGB–IR interaction; and a lightweight SeqShuffleGate (SSG) module that suppresses redundant fusion responses to yield compact and discriminative multimodal representations. Extensive experiments on DroneVehicle, FLIR-Aligned, and LLVIP demonstrate the effectiveness of LDSDet, which achieves 85.2% ${\mathrm{mAP}}_{50}$, 45.3% mAP, and 67.1% mAP, respectively, showing strong robustness under day–night alternation, low-light environments, and complex illumination variations. Full article

Unauthorized commercial drone activity around critical infrastructure motivates the development of trajectory-level anomaly detection. We present a rigorous benchmarking study of variational autoencoder methods for drone trajectory anomaly detection in a simulated nuclear facility protection scenario, evaluating six methods (bidirectional LSTM-VAE, unidirectional LSTM-VAE, fully connected VAE, standard autoencoder, One-Class SVM, Isolation Forest) on 2500 trajectories using identical raw features and training pipelines. Across five random seeds, all VAE variants achieve AUC-ROC of approximately 0.92 versus 0.73 to 0.80 for the non-VAE baselines, isolating variational regularization rather than bidirectionality or temporal encoding alone as the dominant performance driver in this domain. Building on this benchmark, we propose a domain-aware LSTM-VAE incorporating two facility-specific architectural elements: a polar coordinate input representation expressing trajectories relative to the protected facility and a distance-weighted reconstruction loss that allocates model capacity toward near-facility timesteps. The domain-aware variant achieves AUC-ROC of 0.962 ± 0.007 on the original test set and 0.973 ± 0.005 on an augmented hard anomalies test set, a 3 to 4 percentage-point improvement over generic VAE methods at no additional parameter cost. A bootstrap evaluation under 99:1 class imbalance confirms that the domain-aware variant maintains its precision advantage at low false positive rate operating points. Full article

Small surface defects on large structures such as wind turbine blades, bridges, and pipelines pose significant safety risks if left undetected. Therefore, a laser triangulation system is designed for contactless surface geometry inspection from a flying drone at a working distance of 2 m. To enable single-shot triangulation measurements in dynamic aerial environments, a single-shot-capable approach is realized by means of a laser and a diffractive optical element for creating a dot-matrix illumination pattern and a camera for image recording. The setup, with 101 × 101 measurement points, is calibrated by using an interferometer as a reference, which shows a sub-pixel resolution capability. As a result, the depth resolution capability for each point amounts to 126 µm, while the lateral resolution capability is determined by the laser spots’ size of 0.6 mm and the spots’ interspacing of 1.75 mm. With the present configuration, unambiguous depth detection is possible for local surface gradients of up to 2.3 times the interspot distance between adjacent measurement points, and the field of view is 17.56 cm × 17.56 cm. Finally, surface defects with lateral sizes on the order of 1 cm and 0.5 cm are currently detectable, as is demonstrated by experimental results from in-flight measurements. Thus, the potential and challenges of single-shot laser triangulation for drone-based inspection in real-world scenarios are presented. Full article

Methane (CH₄) is a highly potent greenhouse gas whose accurate detection and quantification are essential for climate mitigation and compliance with emerging environmental regulations. Conventional monitoring approaches, including fixed monitoring stations and satellite-based observations, often exhibit limitations in terms of spatial resolution, operational flexibility, and accessibility for localized measurements. This paper presents CH4SCOUT, a modular unmanned aerial vehicle (UAV)-based platform designed for methane detection, environmental monitoring, and georeferenced data acquisition. The proposed system integrates a methane sensing module, environmental sensors, controlled airflow sampling, onboard data acquisition, and wireless communication capabilities within a UAV-compatible architecture. A three-stage signal-conditioning pipeline based on Median filtering, Hampel outlier suppression, and Exponential Moving Average (EMA) smoothing is implemented to improve measurement stability under dynamic flight conditions. Initial real-world validation flights demonstrate stable methane concentration measurements under realistic environmental conditions while maintaining reliable data transmission and telemetry synchronization. Results indicate that low-cost UAV-assisted sensing architectures can provide operationally useful methane measurements when supported by appropriate calibration and deterministic signal conditioning. Future work will focus on advanced plume localization algorithms, autonomous navigation strategies, and enhanced methane emission quantification capabilities. Full article

Object detection in wide-field-of-view (wide-FoV) unmanned aerial vehicle (UAV) imagery for Internet of Things (IoT) remote sensing applications requires accurate recognition of tiny objects under severe background redundancy and extreme scale variation. As the field of view expands, conventional dense detectors tend to waste substantial computation on non-informative regions, while feature downsampling and static receptive fields often cause the dilution of foreground information and scale confusion. To address these issues, we propose MFRC-Det, a unified framework built upon two complementary principles: mask-guided feature routing and adaptive context modeling. Specifically, a Superpixel-Masking Generator (SP-Masker) is introduced to estimate an image-space soft foreground prior by comparing Simple Linear Iterative Clustering (SLIC) superpixel histograms with a peripheral background reference, propagating the resulting scores on a superpixel adjacency graph, and projecting the refined region-level scores back to a pixel-level routing mask. Guided by these priors, a Greedy-Cutter (G-Cutter) converts dense feature maps into compact, foreground-focused patches without repeated backbone evaluation on cropped image regions, thereby reducing redundant background computation while preserving local structural coherence. On top of the retained regions, an Adaptive Receptive-field Selection Network (ARSNet) aggregates multi-scale contextual responses from several learnable receptive-field candidate branches. ARSNet predicts spatial selection weights conditioned on the input features, allowing each location to emphasize a suitable receptive-field response for object representation. Experimental results on VisDrone-DET and UAVDT demonstrate that MFRC-Det achieves competitive detection accuracy with favorable computational efficiency. Specifically, MFRC-Det obtains 36.1% AP, 60.4% $A{P}_{50}$, and 38.5 FPS on VisDrone-DET and 21.3% AP, 36.8% $A{P}_{50}$, and 37.4 FPS on UAVDT. These results validate the effectiveness of mask-guided feature routing and adaptive context modeling for wide-FoV UAV object detection and suggest their potential value for computation-efficient aerial perception in IoT remote sensing applications. Full article

Object detection in aerial imagery faces extreme target sparsity and high-intensity environmental interference, causing weak targets to be submerged in background clutter. To address this, we propose a Collaborative Feature Purification Detection Transformer (CFP-DETR), which reconstructs discriminative target representations through a collaborative feature purification mechanism. Specifically, the Global Context Denoising Module (GCDM) first suppresses environmental noise at the semantic level to enhance target saliency. The purified features are then fused across scales through an Adaptive Cross-scale Feature Alignment (ACFA) module, which resolves spatial misalignment that otherwise dilutes small-object features during multi-level interaction. Concurrently, a Fine-Grained Detail Injection Module (FGDIM) recovers shallow high-resolution details and injects them into the semantic flow, compensating for information loss caused by progressive downsampling. Together, these modules denoise, align, and recover features to counteract submergence at different stages. Additionally, an efficient lightweight variant, Efficient Lightweight CFP-DETR (EL-CFP-DETR), reconstructs the backbone with partial convolution and structural re-parameterization to improve efficiency while maintaining competitive detection accuracy. Extensive experiments across five datasets validate the effectiveness of this collaborative design. On the SeaDronesSee dataset, CFP-DETR increases $A{P}_{50}$ and $A{P}_S^{val}$ by 1.64% and 4.03% over the baseline, while EL-CFP-DETR reduces parameters by 18% to 16.4M and GFLOPs by 15% to 48.3, reaching 42.8 FPS. Notably, CFP-DETR achieves an inference speed of 37.72 FPS, a 31.2% improvement over the baseline Real-Time Detection Transformer (RT-DETR). Full article

To ensure the reliability and sustained performance of industrial photovoltaic (PV) systems, fault detection frameworks must achieve both high detection accuracy and computational efficiency, particularly for deployment on resource-constrained edge platforms. This work proposes a lightweight and low-latency photovoltaic defect detection framework that integrates DCGAN-based generative augmentation with the proposed GhostViT-YOLOv10n architecture. The augmentation strategy helps address class imbalance, improve representation of rare defects, and enhance generalization capability in electroluminescence (EL) imagery through structured geometric and photometric transformations. The proposed framework integrates lightweight Ghost-based optimization, Cross-Stage Partial Fusion (C2f), Spatial Pyramid Pooling—Fast (SPPF), MobileViT contextual learning, and SimAM-based attention refinement to improve multi-scale feature extraction while maintaining low computational complexity. Experimental evaluation on the PVEL-AD and PV Multi Defect benchmark datasets demonstrates strong detection performance. On the PVEL-AD dataset, the BaseLine achieves a mAP@0.5 of 71.6% with only 2.7 M parameters and 8.4 GFLOPs, while our proposed GhostViT-YOLOv10n framework with DCGAN-enhanced version further improves detection performance to 93.6% mAP@0.5 with only 2.19 M parameters and 6.6 GFLOPs. On the PV Multi Defect dataset, the BaseLine achieves a mAP@0.5 of 74.0% with 2.71 M parameters and 8.4 GFLOPs, and the optimized framework with DCGAN-augmented configuration further improves performance to 95.4% mAP@0.5 with 2.58 M parameters and 7.7 GFLOPs. These results demonstrate the effectiveness of combining lightweight architectural optimization with generative augmentation for improving rare defect representation and multi-scale photovoltaic defect detection. To validate practical deployment feasibility, the optimized framework was deployed on a Raspberry Pi 5 using ONNX Runtime under CPU-only conditions. The deployed model achieved an average inference time of 43.05 ms and a real-time processing speed of 23.23 FPS while maintaining moderate CPU utilization and stable thermal behavior. These deployment results demonstrate the suitability of the proposed framework for lightweight edge-oriented photovoltaic inspection applications without requiring GPU acceleration. All evaluations were conducted exclusively on real test datasets, while synthetic samples were used only during training to improve data diversity and rare defect representation. Overall, the proposed framework provides a balanced solution that combines detection accuracy, computational efficiency, lightweight edge deployment capability, and generative augmentation for practical photovoltaic defect inspection applications with potential suitability for future drone-assisted inspection scenarios. Full article

This study examines the influence of roundabout geometric design on motorcyclists’ risky behaviours in mixed-traffic environments, where motorcycles form a dominant traffic component. While conventional safety analyses often emphasise the performance of four-wheeled vehicles, limited attention has been given to how geometric design shapes riders’ behavioural responses and risk perception. To address this gap, this study integrates multiple linear regression and Chi-squared automatic interaction detection (CHAID) to capture both linear effects and nonlinear threshold behaviours. Data were collected from multi-lane roundabouts using a drone and ground-level video observations. Regression results indicate that larger radii and wider geometries are associated with increased risky behaviours, including unsafe stopping and signalling non-compliance. In contrast, CHAID identifies exit radius as the most influential factor, with specific ranges (≤12.43 m and 30.2–32.57 m) associated with more consistent behavioural patterns, while larger radii (>32.57 m) are linked to increased risk. These findings highlight the importance of context-sensitive, motorcycle-oriented geometric design in improving safety outcomes in mixed-traffic environments. Full article

UAV-based wildlife monitoring encounters tremendous challenges posed by complex environments, such as the extremely low proportion of effective targets in aerial images and variations in remote sensing scales. This paper presents a novel fusion framework named YOLIP, which integrates a detection head with semantic perception capabilities and an implicit feature adjustment module to boost detection accuracy and feature representation ability. Specifically, this paper redesigns the detection head to enable it to simultaneously learn spatial positioning and semantic features, thereby achieving more reliable extraction of regional features. The implicit feature modulation module introduces a dual-path fusion mechanism, which elevates the feature quality through geometric–semantic fusion, thereby improving the consistency and robustness of the detection. Furthermore, this paper also develops an asynchronous scheduling strategy, which can selectively execute computationally intensive operations to achieve computational optimization, enabling this framework to adapt to actual detection scenarios based on unmanned aerial vehicles. In this study, we conducted numerous experiments on the self-built drone wildlife dataset as well as the publicly available aerial wildlife dataset. Theresults demonstrate that compared with existing detection models, YOLIP improves mAP@0.5 by 11.6% while maintaining an efficient inference speed, achieving an improvement in detection performance. In addition, cross-dataset evaluation verifies the stable performance and generalization capability of the proposed method across multiple real-world scenarios. Full article

The rapidly increasing availability of low-cost commercial UAVs poses significant security challenges for critical infrastructure and law enforcement agencies. This paper presents an integrated LiDAR-based detection and vision-based verification system for an autonomous drone-on-drone aerial interception system. To eliminate the threat of possible dangerous target drones, the interception UAVs presented in this paper use a net to capture them safely in the air. The system addresses the critical limitation of ground-based sensors, which provide insufficient precision for reliable net-based capture operations. Moving beyond simulation-only approaches, the core novelty of this work lies in the successful real-world integration of these sensors on a strictly constrained aerial platform in size, weight and power to achieve sub-meter terminal guidance precision. The developed system uses real-time point cloud processing, DBSCAN clustering, and Moving Horizon Estimation tracking for the detection and tracking of the target. Vision-based verification uses a custom-trained YOLO neural network and achieves over 90% detection rates. The evaluation demonstrates a detection accuracy of less than 0.4 m at ranges exceeding 40 m during dynamic interception scenarios using RTK-GNSS ground truth. The dual-sensor approach successfully completed multiple autonomous interception missions with target detection ranges of up to 60 m, validating the capability of the system for safe, autonomous civilian UAV interception. Full article