Search Results (2,225)

Accurate real-time detection of blueberry maturity is vital for automated harvesting. However, existing methods often fail under occlusion, variable lighting, and dense fruit distribution, leading to reduced accuracy and efficiency. To address these challenges, we designed a lightweight deep learning framework that integrates improved feature extraction, attention-based fusion, and progressive transfer learning to enhance robustness and adaptability To overcome these challenges, we propose BMDNet-YOLO, a lightweight model based on an enhanced YOLOv8n. The backbone incorporates a FasterPW module with parallel convolution and point-wise weighting to improve feature extraction efficiency and robustness. A coordinate attention (CA) mechanism in the neck enhances spatial-channel feature selection, while adaptive weighted concatenation ensures efficient multi-scale fusion. The detection head employs a heterogeneous lightweight structure combining group and depthwise separable convolutions to minimize parameter redundancy and boost inference speed. Additionally, a three-stage transfer learning framework (source-domain pretraining, cross-domain adaptation, and target-domain fine-tuning) improves generalization. Experiments on 8,250 field-collected and augmented images show BMDNet-YOLO achieves 95.6% mAP@0.5, 98.27% precision, and 94.36% recall, surpassing existing baselines. This work offers a robust solution for deploying automated blueberry harvesting systems. Full article

The growing demand for reliable long-term traffic forecasting has become increasingly critical in the development of intelligent transportation systems (ITS). However, capturing both strong periodic patterns and long-range temporal dependencies presents a significant challenge, and existing approaches often fail to balance these factors effectively, resulting in unstable or suboptimal predictions. To address this issue, we propose MixModel , a novel hybrid framework that integrates TimesNet and Informer to leverage their complementary strengths. Specifically, the TimesNet branch extracts periodic variations through frequency-domain decomposition and multi-scale convolution, while the Informer branch employs ProbSparse attention to efficiently capture long-range dependencies across extended horizons. By unifying these capabilities, MixModel achieves enhanced forecasting accuracy, robustness, and stability compared with state-of-the-art baselines. Extensive experiments on real-world highway datasets demonstrate the effectiveness of our model, highlighting its potential for advancing large-scale urban traffic management and planning. To the best of our knowledge, MixModel is the first hybrid framework that explicitly bridges frequency-domain periodic modeling and efficient long-range dependency learning for long-term traffic forecasting, establishing a new benchmark for future research in Intelligent Transportation Systems. Full article

Reduced visibility caused by foggy weather has a significant impact on transportation systems and driving safety, leading to increased accident risks and decreased operational efficiency. Traditional methods rely on expensive physical instruments, limiting their scalability. To address this challenge in a cost-effective manner, we propose a two-stage network for visibility estimation from stereo image inputs. The first stage computes scene depth via stereo matching, while the second stage fuses depth and texture information to estimate metric-scale visibility. Our method produces pixel-wise visibility maps through a physically constrained, progressive supervision strategy, providing rich spatial visibility distributions beyond a single global value. Moreover, it enables the detection of patchy fog, allowing a more comprehensive understanding of complex atmospheric conditions. To facilitate training and evaluation, we propose an automatic fog-aware data generation pipeline that incorporates both synthetically rendered foggy images and real-world captures. Furthermore, we construct a large-scale dataset encompassing diverse scenarios. Extensive experiments demonstrate that our method achieves state-of-the-art performance in both visibility estimation and patchy fog detection. Full article

The advent of sixth-generation (6G) wireless networks and cell-free massive multiple-input multiple-output (MIMO) architectures underscores the need for efficient resource allocation to support federated learning (FL) at the network edge. Existing approaches often treat communication, computation, and learning in isolation, overlooking dynamic heterogeneity and fairness, which leads to degraded performance in large-scale deployments. To address this gap, we propose a joint optimization framework that integrates communication–computation co-design, fairness-aware aggregation, and a hybrid strategy combining convex relaxation with deep reinforcement learning. Extensive experiments on benchmark vision datasets and real-world wireless traces demonstrate that the framework achieves up to 23% higher accuracy, 18% lower latency, and 21% energy savings compared with state-of-the-art baselines. These findings advance joint optimization in federated learning (FL) and demonstrate scalability for 6G applications. Full article

Current deep learning-based dense target detection models face dual challenges in industrial scenarios: high computational complexity leading to insufficient inference efficiency on mobile devices, and missed/false detections caused by dense small targets, high inter-class similarity, and complex background interference. To address these issues, this paper proposes DBA-YOLO, a lightweight model based on YOLOv10, which significantly reduces computational complexity through model compression and algorithm optimization while maintaining high accuracy. Key improvements include the following: (1) a C2f PA module for enhanced feature extraction, (2) a parameter-refined BIMAFPN neck structure to improve small target detection, and (3) a DyDHead module integrating scale, space, and task awareness for spatial feature weighting. To validate DBA-YOLO, we constructed a real-world dataset from cigarette package images. Experiments on SKU-110K and our dataset show that DBA-YOLO achieves 91.3% detection accuracy (1.4% higher than baseline), with mAP and mAP75 improvements of 2–3%. Additionally, the model reduces parameters by 3.6%, balancing efficiency and performance for resource-constrained devices. Full article

Efficient simplification of 3D models is essential for mobile and other resource-constrained application scenarios. Industrial 3D assemblies, typically composed of numerous components and dense triangular meshes, often pose significant challenges in rendering and transmission due to their large scale and high complexity. The Quadric Error Metrics (QEM) algorithm offers a practical balance between simplification accuracy and computational efficiency. However, its application to large-scale industrial models remain limited by performance bottlenecks, especially when combined with curvature-based optimization techniques that improve fidelity at the cost of increased computation. Therefore, this paper presents a parallel implementation of the QEM algorithm and its curvature-optimized variant using the OpenMP framework. By identifying key bottlenecks in the serial workflow, this research parallelizes critical processes such as curvature estimation, error metric computation, and data structure manipulation. Experiments on large industrial assembly models at a simplification ratio of 0.3, 0.5, and 0.7 demonstrate that the proposed parallel algorithms achieve significant speedups, with a maximum observed speedup of 5.5×, while maintaining geometric quality and topological consistency. The proposed approach significantly improves model processing efficiency, particularly for medium- to large-scale industrial models, and provides a scalable and practical solution for real-time loading and interaction in engineering applications. Full article

Deep learning-based Non-Intrusive Load Monitoring (NILM) methods have been widely applied to residential load identification. However, photovoltaic (PV) loads exhibit strong non-stationarity, high dependence on weather conditions, and strong coupling with multi-source data, which limit the accuracy and generalization of existing models. To address these challenges, this paper proposes a multi-scale and multi-feature fusion framework for PV disaggregation, consisting of three modules: Multi-Scale Time Series Decomposition (MTD), Multi-Feature Fusion (MFF), and Temporal Attention Decomposition (TAD). These modules jointly capture short-term fluctuations, long-term trends, and deep dependencies across multi-source features. Experiments were conducted on real residential datasets from southern China. Results show that, compared with representative baselines such as SGN-Conv and MAT-Conv, the proposed method reduces MAE by over 60% and SAE by nearly 70% for some users, and it achieves more than 45% error reduction in cross-user tests. These findings demonstrate that the proposed approach significantly enhances both accuracy and generalization in PV load disaggregation. Full article

The paper presents the main features of a Web-GIS platform designed to compute real-time scenario-based seismic risk assessments at the national level. Based on the Italian experience, the platform enables DRM scientist and policymakers to readily generate seismic scenarios supporting the entire DRM cycle, including training, emergency planning, calibrating operations during response, and providing seismic risk estimates for National Disaster Risk Assessment or seismic risk reduction programs. The platform is immediately operational, relying on preloaded freeware datasets on exposure and vulnerability, and requiring only basic earthquake parameters to perform real-time analysis. At a later stage, these datasets should be replaced with more detailed and accurate national-level data. The platform generates earthquake impact assessments that include physical damage, economic and human losses, and key emergency response indicators, such as estimated displaced population, required tent camps, and EMT and USAR needs. Its key innovation lies in the ability to operate at the national scale, offering immediate usability with the possibility of further customization. As a web-based service with a user-friendly graphical interface, it is particularly suited for civil protection and DRM experts. Full article

To address the issues of small-target missed detection, false alarms from cloud/fog interference, and low computational efficiency in traditional wildfire detection for transmission line corridors, this paper proposes a YOLOv11_MDS detection model by integrating Multi-Scale Convolutional Attention (MSCA) and Distribution-Shifted Convolution (DSConv). The MSCA module is embedded in the backbone and neck to enhance multi-scale dynamic feature extraction of flame and smoke through collaborative depth strip convolution and channel attention. The DSConv with a quantized dynamic shift mechanism is introduced to significantly reduce computational complexity while maintaining detection accuracy. The improved model, as shown in experiments, achieves an mAP@0.5 of 88.21%, which is 2.93 percentage points higher than the original YOLOv11. It also demonstrates a 3.33% increase in recall and a frame rate of 242 FPS, with notable improvements in detecting small targets (pixel occupancy < 1%). Generalization tests demonstrate mAP improvements of 0.4% and 0.7% on benchmark datasets, effectively resolving false/missed detection in complex backgrounds. This study provides an engineering solution for real-time wildfire monitoring in transmission lines with balanced accuracy and efficiency. Full article

We present a (Light)weight (S)atellite–(A)erial feature (M)atching framework (LiteSAM) for robust UAV absolute visual localization (AVL) in GPS-denied environments. Existing satellite–aerial matching methods struggle with large appearance variations, texture-scarce regions, and limited efficiency for real-time UAV applications. LiteSAM integrates three key components to address these issues. First, efficient multi-scale feature extraction optimizes representation, reducing inference latency for edge devices. Second, a Token Aggregation–Interaction Transformer (TAIFormer) with a convolutional token mixer (CTM) models inter- and intra-image correlations, enabling robust global–local feature fusion. Third, a MinGRU-based dynamic subpixel refinement module adaptively learns spatial offsets, enhancing subpixel-level matching accuracy and cross-scenario generalization. The experiments show that LiteSAM achieves competitive performance across multiple datasets. On UAV-VisLoc, LiteSAM attains an RMSE@30 of 17.86 m, outperforming state-of-the-art semi-dense methods such as EfficientLoFTR. Its optimized variant, LiteSAM (opt., without dual softmax), delivers inference times of 61.98 ms on standard GPUs and 497.49 ms on NVIDIA Jetson AGX Orin, which are 22.9% and 19.8% faster than EfficientLoFTR (opt.), respectively. With 6.31M parameters, which is 2.4× fewer than EfficientLoFTR’s 15.05M, LiteSAM proves to be suitable for edge deployment. Extensive evaluations on natural image matching and downstream vision tasks confirm its superior accuracy and efficiency for general feature matching. Full article

Detecting surface cracks on sugarcane is a critical step in ensuring product quality control, with detection precision directly impacting raw material screening efficiency and economic benefits in the sugar industry. Traditional methods face three core challenges: (1) complex background interference complicates texture feature extraction; (2) variable crack scales limit models’ cross-scale feature generalization capabilities; and (3) high computational complexity hinders deployment on edge devices. To address these issues, this study proposes a lightweight sugarcane surface crack detection model, SCS-YOLO (Surface Cracks on Sugarcane-YOLO), based on the YOLOv10 architecture. This model incorporates three key technical innovations. First, the designed RFAC2f module (Receptive-Field Attentive CSP Bottleneck with Dual Convolution) significantly enhances feature representation capabilities in complex backgrounds through dynamic receptive field modeling and multi-branch feature processing/fusion mechanisms. Second, the proposed DSA module (Dynamic SimAM Attention) achieves adaptive spatial optimization of cross-layer crack features by integrating dynamic weight allocation strategies with parameter-free spatial attention mechanisms. Finally, the DyHead detection head employs a dynamic feature optimization mechanism to reduce parameter count and computational complexity. Experiments demonstrate that on the Sugarcane Crack Dataset v3.1, compared to the baseline model YOLOv10, our model achieves mAP50:95 to 71.8% (up 2.1%). Simultaneously, it achieves significant reductions in parameter count (down 19.67%) and computational load (down 11.76%), while boosting FPS to 122 to meet real-time detection requirements. Considering the multiple dimensions of precision indicators, complexity indicators, and FPS comprehensively, the SCS—YOLO detection framework proposed in this study provides a feasible technical reference for the intelligent detection of sugarcane quality in the raw materials of the sugar industry. Full article

Decarbonisation of urban energy and transportation systems has become a priority for cities worldwide, with policies primarily aiming to promote rooftop solar electricity generation and a shift to private electric vehicles (EVs). However, these policies may also increase inequalities in access to affordable, low-carbon mobility and the associated benefits. While academic literature shows increasing awareness of these equity impacts and risks, the extent to which this applies to policy practice remains unclear. We therefore conducted a case study of seven Dutch cities, analysing local policy documents and conducting interviews with policymakers. The study provided insight into the current policy landscape and revealed a general sensitivity among interviewed policymakers to possible equity impacts of the current decarbonisation policies. Only a few measures to address these impacts are currently in place, but policymakers have proposed a range of novel and more inclusive measures, which can be tested for their impacts and scaling potential in real-life experiments. Another priority for future research is exploring the potential of shared electric mobility to provide equitable access to low-carbon transportation. Full article

Multi-Robot Task Assignment (MRTA) is a critical and inherently multi-objective problem in diverse real-world applications, demanding the simultaneous optimization of conflicting objectives such as minimizing total travel distance and balancing robot workload. Existing multi-objective evolutionary algorithms (MOEAs) often struggle with slow convergence and insufficient diversity when tackling the combinatorial complexity of large-scale MRTA instances. This paper introduces the Collaborative Swarm-Differential Evolution (CSDE) algorithm, a novel MOEA designed to overcome these limitations. CSDE’s core innovation lies in its deep, operator-level fusion of Differential Evolution’s (DE) robust global exploration capabilities with Particle Swarm Optimization’s (PSO) swift local exploitation prowess. This is achieved through a unique fused velocity update mechanism, enabling particles to dynamically benefit from their personal experience, collective swarm intelligence, and population diversity-driven knowledge transfer. Comprehensive experiments on various MRTA scenarios demonstrate that CSDE consistently achieves superior performance in terms of convergence, solution diversity, and Pareto front quality, significantly outperforming standard multi-objective algorithms like Multi-Objective Particle Swarm Optimization (MOPSO), Multi-Objective Differential Evolution (MODE), and Multi-Objective Genetic Algorithm (MOGA). This study highlights CSDE’s substantial contribution to the MRTA field and its potential for more effective and efficient multi-robot system deployment. Full article

High-dimensional time series data forecasting has been a popular problem in recent years, with ubiquitous applications in both scientific and business fields. Modern datasets may incorporate thousands of correlated time series that evolve together, and correctly identifying the correlated patterns and modeling the inter-series relationship can significantly promote forecast accuracy. However, most statistical methods are inadequate for handling complicated time series due to violation of model assumptions, and most recent deep learning approaches in the literature are either univariate (not fully utilizing inter-series information) or computationally expensive. This paper present SM-TCN, a Sparse Multi-scale Temporal Convolutional Network, utilizing a forward–backward residual architecture with sparse TCN kernels of different lengths to extract multi-resolution characteristics, which sufficiently reduces computational complexity specifically for high-dimensional problems. Extensive experiments on real-world datasets have demonstrated that SM-TCN outperforms state-of-the-art approaches by 10% in MAE and MAPE, and has the additional advantage of high computation efficiency. Full article

Transparent objects are prevalent in various everyday scenarios. However, their reflective and refractive optical properties present significant challenges for conventional optical sensors. This difficulty makes the task of generating dense depth maps from sparse depth maps and high-resolution RGB images a critical area of research. In this paper, we introduce SRNet-Trans, a novel two-stage depth completion framework specifically designed for transparent objects. The approach is structured into two stages, each primarily focused on leveraging semantic and depth information, respectively. In the first stage, RGB images and sparse depth maps are used to predict a relatively dense depth map. The second stage then takes the predicted depth from the first stage, along with the sparse depth map, to generate a final dense depth map. The depth information produced by the two stages is complementary, allowing for effective fusion of both outputs. To enhance the depth estimation process, we integrate a self-attention mechanism in the first stage to better capture semantic features and introduce geometric convolutional layers in the second stage to improve depth encoding accuracy. Additionally, we incorporate a global consistency-based fine depth recovery technique to further refine the final depth map. Extensive experiments on the large-scale real-world TransCG dataset demonstrate that SRNet-Trans outperforms current state-of-the-art methods in terms of depth estimation accuracy. Full article