Search Results (66)

Physical-domain adversarial attacks have been extensively studied in face recognition and object detection, yet the field of palmprint recognition remains largely unexplored. Existing methods generate grayscale patches constrained by the single-channel input of most palmprint models. When deployed on skin, these patches contrast sharply with the surrounding tissue and are readily noticeable to human observers, undermining the covertness required in practical attacks. To address this limitation, we propose a Color Adversarial Patch (CAP) generation algorithm that leverages style transfer principles to produce visually natural color patches while maintaining high attack success rates. The method initiates the patch with a style prior using a pre-trained Contrastive Arbitrary Style Transfer (CAST) model and jointly optimizes adversarial loss, style loss, and smoothness loss within a unified framework. A three-channel averaging strategy is adopted to ensure compatibility with single-channel recognition models during gradient backpropagation. Experiments on the Tongji palmprint dataset show that the generated color patches achieve average cosine similarity values above the decision threshold in physical-domain tests, with peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) values significantly higher than those for their grayscale counterparts. Ablation studies validate the indispensable role of each loss component. CAP offers a practical balance between attack effectiveness and visual camouflage, demonstrating the feasibility of concealed physical-domain attacks on palmprint recognition systems. Full article

The detection of cracks in concrete bridges is essential for evaluating structural durability and load-carrying capacity. Although vision-based crack detection methods have been widely studied, the detection of fine cracks remains challenging. This study proposes a method for detecting fine surface cracks in bridges based on close-range unmanned aerial vehicle (UAV) photography and a camouflaged object detection network. The main contributions are as follows: 1. The feasibility of using a UAV equipped with a telephoto camera to capture fine cracks was investigated, and close-range imaging was shown to enable the acquisition of fine cracks as narrow as 0.08 mm. 2. To address the difficulty of accurately identifying cracks on concrete bridge surfaces due to stain interference and the weak texture features of fine cracks, a crack segmentation method based on a boundary-guided camouflaged object detection network was applied and validated. By improving the contextual aggregation module, the method achieved accurate identification of fine cracks under stain interference. 3. To overcome the difficulty of accurately measuring crack width when fine cracks occupy only a small number of pixels, a deep learning-based super-resolution method was applied to achieve sub-pixel-level crack width measurement. The proposed method was tested and validated on an in-service concrete bridge. The test results show that the proposed method achieved a mean absolute error of 0.7% in crack segmentation and a width measurement error of less than 0.06 mm, demonstrating its practical applicability. Full article

Camouflaged object detection (COD) remains a challenging task because objects blend into the background, exhibiting low contrast, incomplete edges, and highly similar appearances. Recent deep learning methods have improved detection performance, but most rely solely on visual features and lack semantic-level reasoning to distinguish concealed objects. To address both performance and deployment scalability, we use an edge–cloud network paradigm in which lightweight visual processing operates on an internet device while semantic reasoning is handled remotely, enabling real-time COD in internet-scale applications such as wildlife monitoring, perimeter surveillance, and UAV-based sensing. We implement this in SPINet, a vision–language-driven hybrid COD framework that integrates a multi-stage BiRefNet-Large decoder with BLIP-Large semantic comprehension. Our Multi-Stage Semantic Prior Injection (MS-SPI) module extracts and injects three complementary semantic representations for global context, region-level features, and spatial attention maps into three decoder stages (Stages 3, 4, and 5) of BiRefNet, enabling hierarchical semantic guidance at multiple scales. Experiments on three benchmark datasets (COD10K, CAMO, and NC4K) demonstrate that SPINet achieves consistent improvements over the BiRefNet-Large visual-only baseline across all benchmarks. SPINet attains $S_{\alpha }=0.921$ on COD10K, $0.859$ on CAMO, and $0.893$ on NC4K, outperforming BiRefNet-Large by $+0.9\%$, $+1.7\%$, and $+3.5\%$ in structure measure, respectively, with MAE reductions of $7\%$, $29\%$, and $22\%$. These results show that frozen semantic priors provide robust and transferable guidance for COD with negligible additional parameters. Full article

Underwater object detection is crucial for marine ecological monitoring and resource exploration. However, low underwater contrast causes severe aliasing between foreground and background in the spatial domain, and conventional methods struggle to effectively decouple their features. Transformer architectures incur high computational overhead. Conversely, over-compressing these models severely degrades their ability to detect heavily camouflaged or small marine organisms. High-quality underwater samples are limited, and existing time-consuming generative strategies are hard to implement efficiently on edge devices. To address these challenges, this paper proposes a resource-efficient framework for underwater object detection. First, we design a Wavelet-Enhanced Feature Pyramid Network that combines a saliency-focus mechanism and a discrete wavelet transform to overcome background noise in both spatial and frequency domains, extracting features of hidden small objects. Second, a data-dependent dynamic token pruning technique removes redundant tokens, effectively mitigating the computational bottleneck without sacrificing essential semantic capacity. Finally, for extreme sample scarcity, we introduce a Feature Correction Module and a two-stage fine-tuning and feature correction strategy, using a high-precision teacher model to guide a compressed student network in adaptively compensating for optical shifts with few samples. Experiments on URPC2020 and DUO demonstrate that our method improves small object detection accuracy while reducing parameter count and computational overhead, striking a good balance between accuracy and inference efficiency. Full article

The development of transparent multifunctional coatings capable of combining optical properties with mechanical durability remains a significant challenge in advanced materials engineering. In this study, novel hybrid coatings based on a poly(vinyl butyral)/epoxy resin (PVB/epoxy) matrix reinforced with tungsten disulfide (WS₂) nanostructures were developed and examined for potential application in camouflage protection of glass surfaces. Camouflage aims to reduce the detectability of an object by minimizing the optical contrast between the object and its surrounding environment. For transparent substrates such as glass, this objective is particularly demanding because the transparency must be preserved while reducing unwanted surface reflection and optical signatures over relevant spectral ranges. For this purpose, in this research two types of nanostructures were investigated: fullerene-like nanoparticles (IF-WS₂) and inorganic nanotubes (INT-WS₂). The coatings were fabricated via ultrasonically assisted solution dispersion followed by casting over the glass plates and Teflon molds, and solvent evaporation. Structural, thermal, optical, and mechanical properties were systematically evaluated using SEM, FTIR, DSC, UV-Vis-NIR spectroscopy, gloss measurements, hardness testing, and cavitation wear resistance analysis. The incorporation of WS₂ nanostructures led to improved mechanical performance, with increased hardness and enhanced resistance to cavitation-induced wear. Optical characterization showed moderate reductions in reflectance and controlled transmittance in the visible and near-infrared regions, while overall transparency was maintained. The results indicate that WS₂ nanostructures contribute to both light scattering and absorption, leading to reduced specular reflection and improved optical masking potential. The findings demonstrate that hybrid PVB/epoxy/WS₂ coatings offer a promising approach for designing transparent, mechanically resistant coatings with tunable optical properties, with potential applications in protective glass systems and advanced functional surfaces. Full article

Camouflaged object detection (COD) methods suffer severe degradation on small camouflaged objects and remain incapable of real-time inference on edge devices. We propose LeanCOD, a framework that pairs a strong foundation-model encoder with a lightweight decoder to enable high-resolution inference. A size-aware composite loss further strengthens supervision on small camouflaged objects. Our size-wise experiments reveal that the 0–1% extra-small-object regime is the major performance bottleneck for existing COD methods. LeanCOD achieves an $S_{\alpha }$ of 0.915 on COD10K at a $576\times 576$ resolution, outperforming competing methods at equal or lower resolutions. Deployed with TensorRT FP16 on an NVIDIA Jetson AGX Orin, LeanCOD runs at 31.6 FPS while maintaining an $S_{\alpha }$ of 0.908 at a $576\times 576$ resolution, exceeding the 30 FPS real-time threshold. Full article

Adversarial camouflage has attracted growing research attention owing to its ability to execute multi-view, persistent attacks in real physical environments, outperforming conventional single-view adversarial patches. However, most existing methods are confined to non-targeted attacks, which induce arbitrary incorrect detection results without specifying target categories. This ambiguity weakens attack destructiveness and stealthiness, posing limitations for security evaluation of real-world vision systems. To address this gap, we present TACT, an approach built upon the full-coverage physical camouflage pipeline. By replacing the original category supervision with a predefined target class, TACT redirects the optimization gradient to guide 3D texture toward the target category features. Such a scheme only employs the inherent feature alignment mechanism of off-the-shelf object detectors, without redesigning network modules, defining novel loss functions, or modifying the rendering pipeline. Extensive experiments across digital and physical domains validate its effectiveness: on seven mainstream general-purpose object detectors, TACT-person achieves an average targeted attack success rate of 51.91%, and delivers cross-architecture and cross-version transferability. In physical tests, TACT-bird reduces mAP50-95 by 59.87% on YOLOv8, yet a TCER–TASR gap suggests that the physical pipeline acts as a low-pass filter: coarse-grained target classes transfer robustly while fine-grained ones suffer feature collapse. These results confirm the viability of native supervision redirection and reveal an empirical pattern: coarse-grained target classes transfer more robustly through the physical pipeline than fine-grained ones, suggesting that target class feature granularity consistently influences physical-domain attack effectiveness. Full article

Camouflaged object detection (COD) is important for intelligent UAV monitoring and search-and-rescue operations. However, existing benchmarks focus primarily on natural camouflage, creating a noticeable domain shift for specific applications such as the search and rescue of individuals visually similar to their surroundings due to their clothing. To investigate this shift, we introduce CPD-UAV, a benchmark comprising 1061 high-resolution images with detailed pixel-level annotations across diverse terrains and flight altitudes. Benchmarking of seven state-of-the-art models on this dataset reveals specific challenges. Specifically, the scale variations and “vanishing boundaries” inherent in aerial perspectives can lead to boundary localization inaccuracies. Furthermore, this evaluation observes the deceptive nature of traditional metrics, such as Mean Absolute Error (MAE), when targets occupy small image proportions. To address the degradation of weak target signals during feature integration, we propose a lightweight, plug-and-play component: the Residual Gated Alignment Module (RGAM). RGAM handles scale variations by establishing semantic anchors in deep network layers, mitigating signal dilution and highlighting micro-targets against complex backgrounds. By integrating RGAM into three representative baselines, we demonstrate that the enhanced architectures achieve a competitive performance level. Quantitative results show consistent improvements in structural integrity (structure-measure, $S_m$) and boundary localization. Ultimately, this work provides a practical data platform and an effective algorithmic solution for advancing aerial monitoring systems. Full article

Welfare assessment for the endangered red panda (Ailurus fulgens) in captivity requires systematic behaviour monitoring, yet traditional direct observation is often limited by observer subjectivity and time constraints. This study evaluates the feasibility of employing machine learning (ML) to automate behavioural monitoring of a red panda in a complex, mixed-species enclosure at Aalborg Zoo, Denmark. Using video data from cameras in the enclosure of the red panda, and the ML model LabGym for animal detection and behavioural categorisation, models were trained to analyse activity patterns of the red panda. The results demonstrate that, while the behaviour categorizer is a promising tool with high classification confidence, the overall system effectiveness is currently limited by the object detector’s performance in a naturalistic environment. Challenges such as environmental obstructions (e.g., rocks, foliage, and trees) and the animal’s camouflage contributed to a significant amount of unclassified time, which may affect the overall assessment of behavioural distribution. We conclude that, while ML holds potential for non-invasive behaviour monitoring, its application in complex zoo settings requires improved detection capabilities to be fully reliable. Future iterations of this system could be enhanced by complementing standard object detection with pose estimation frameworks. Implementing alternative labelling strategies or background subtraction methods could additionally mitigate the detection challenges posed by environmental obstruction. Full article

In aerial sensor systems, detecting helicopters against diverse backgrounds remains challenging due to environmental camouflage. This paper proposes an end-to-end framework for generating adaptive camouflage patterns to evade YOLO-based object detection. Starting with synthetic sensor imagery (background + transparent helicopter overlay), we employ a fine-tuned YOLOv8m for precise VTOL mask extraction, followed by KMeans clustering with Gaussian blur for dominant color extraction from the background. These colors guide Stable Diffusion inpainting to synthesize full-screen camouflage textures, which are then masked and overlapped onto the helicopter region. Evaluated on a 920-image dataset across multiple backgrounds, our method achieves a 97.6% reduction in mAP@0.5 (from 0.8175 to 0.0196) on 751 camouflaged images against a fine-tuned YOLOv8m model, with recall dropping by 95.9%. Even against a helicopter-specialized Defence model, mAP@0.5 drops by 89.6% (from 0.1178 to 0.0123). Ablation studies confirm the synergy of YOLO masking and color-guided inpainting. This sensor-fusion approach enhances stealth in unmanned aerial surveillance, with implications for civilian aviation safety. Full article

The efficacy of Underwater Camouflaged Object Detection (UCOD) is fundamentally constrained by severe boundary ambiguity, where biological mimicry blends targets into complex backgrounds and aquatic optical degradation erodes edge details. We propose a lightweight boundary perception detector named CAR-YOLO (Camouflage Ambiguity Resolution YOLO). Specifically, a frequency-domain dual-path mechanism (FRM-DWT/EG-IWT) leverages selective wavelet aggregation and dynamic injection to recover high-frequency edges. Subsequently, these high-frequency cues are synergized with low-frequency semantic information via the Low-level Adaptive Fusion (LAF) module. To further address noisy samples, an Uncertainty Calibration Head (UCH) refines supervision via prediction consistency. Finally, we constructed specialized datasets based on public data for training and evaluation, including UCOD10K and UWB-COT220. On UCOD10K, CAR-YOLO achieves 27.1% mAP50–95, surpassing several state-of-the-art (SOTA) methods while reducing parameters from 2.58 M to 2.43 M and GFLOPs from 6.3 to 5.9. On the challenging UWB-COT220 benchmark, the model attains 30.7% mAP50–95, marking a 7.7-point improvement over YOLOv11. Furthermore, cross-domain experiments on UODD demonstrate strong generalization. These results indicate that CAR-YOLO effectively mitigates boundary ambiguity, achieving an optimal balance between accuracy, robustness, and efficiency. Full article

The olive psyllid (Euphyllura olivina), previously considered a secondary pest in Spain, is becoming more prevalent due to climate change and rising average temperatures. Its cottony wax secretions can cause substantial damage to olive crops under certain climatic conditions. Traditional monitoring methods for this pest are often labor-intensive, subjective, and impractical for large-scale surveillance. This study presents an automatic image segmentation approach based on deep learning to detect and quantify the cottony masses produced by E. olivina in olive trees. A well-annotated image dataset is developed and published, and a thorough evaluation of current camouflaged object detection (COD) methods is carried out for this task. Our results show that deep learning-based segmentation enables accurate and non-invasive assessment of pest symptoms, even in challenging visual conditions. However, further calibration and field validation are required before these methods can be deployed for operational integrated pest management. This work establishes a public dataset and a baseline benchmark, providing a foundation for future research and decision-support tools in precision agriculture. Full article

Camouflaged Object Detection (COD) is a challenging computer vision task aimed at accurately identifying and segmenting objects seamlessly blended into their backgrounds. This task has broad applications across medical image segmentation, defect detection, agricultural image detection, security monitoring, and scientific research. Traditional COD methods often struggle with precise segmentation due to the high similarity between camouflaged objects and their surroundings. In this study, we introduce a Boundary-Guided Differential Attention Network (BDA-Net) to address these challenges. BDA-Net first extracts boundary features by fusing multi-scale image features and applying channel attention. Subsequently, it employs a differential attention mechanism, guided by these boundary features, to highlight camouflaged objects and suppress background information. The weighted features are then progressively fused to generate accurate camouflage object masks. Experimental results on the COD10K, NC4K, and CAMO datasets demonstrate that BDA-Net outperforms most state-of-the-art COD methods, achieving higher accuracy. Here we show that our approach improves detection accuracy by up to 3.6% on key metrics, offering a robust solution for precise camouflaged object segmentation. Full article

Deep neural networks have demonstrated remarkable performance in object detection tasks; however, they remain highly susceptible to adversarial attacks. Previous surveys in computer vision have provided considerable coverage of physical adversarial attacks, yet the aspect of systematically categorizing and evaluating their physical deployment methods for object detection has not received commensurate focus. To address this gap, we categorize physical adversarial attacks into three primary classes based on the physical deployment of adversarial patterns: manipulating 2D physical objects, injecting adversarial signals, and placing 3D adversarial camouflage. These categories are further analyzed and compared across nine key attributes. Furthermore, we elucidate the relationship between physical adversarial attacks against object detection models and two critical properties: transferability and perceptibility. Our findings indicate that while attacks involving the manipulation of 2D physical objects are relatively straightforward to deploy, their adversarial patterns are often perceptible to human observers. Similarly, 3D adversarial camouflage tends to lack stealthiness, whereas adversarial signal injection offers stronger imperceptibility. However, all three attack types exhibit limited transferability across different models and modalities. Finally, we discuss current challenges and propose actionable directions for future research, aiming to foster the development of more robust object detection systems. Full article

Anti-tank landmines endanger post-conflict recovery. Detecting camouflaged TM-62 landmines in low-altitude unmanned aerial vehicle (UAV) imagery is challenging because targets occupy few pixels and are low-contrast and often occluded. We introduce a single-class anti-tank dataset and a YOLOv5 pipeline augmented with a SAHI-based small-object stage and Weighted Boxes Fusion. The evaluation combines COCO metrics with an operational operating point (score = 0.25; IoU = 0.50) and stratifies by object size and occlusion. On a held-out test partition representative of UAV acquisition, the baseline YOLOv5 attains mAP@0.50:0.95 = 0.553 and AP@0.50 = 0.851. With tuned SAHI (768 px tiles, 40% overlap) plus fusion, performance rises to mAP@0.50:0.95 = 0.685 and AP@0.50 = 0.935—ΔmAP = +0.132 (+23.9% rel.) and ΔAP@0.50 = +0.084 (+9.9% rel.). At the operating point, precision = 0.94 and recall = 0.89 (F1 = 0.914), implying a 58.4% reduction in missed detections versus a non-optimized SAHI baseline and a +14.3 AP@0.50 gain on the small/occluded subset. Ablations attribute gains to tile size, overlap, and fusion, which boost recall on low-pixel, occluded landmines without inflating false positives. The pipeline sustains real-time UAV throughput and supports actionable triage for humanitarian demining, as well as motivating RGB–thermal fusion and cross-season/-domain adaptation. Full article