Search Results (6,769)

Accurate and rapid detection of citrus leaf diseases is important for early diagnosis, precision orchard management, and the reduction of economic losses in citrus production. Automatic detection remains difficult because early lesions are often small and irregular. Several disease categories also share similar visual appearances, and localization is easily affected by veins, shadows, and cluttered backgrounds. To address these task-specific challenges, we propose HGSM-YOLO, where HGSM denotes the coordinated use of heterogeneous convolution, a GSConv-based slim neck, and multi-scale dilated local attention. The framework is built on YOLO11n because its 2.59 M-parameter and 6.4 GFLOP design provides a stringent compact baseline for edge-oriented improvement. The method follows a hierarchical design: C3k2-HetConv preserves lesion edges and local morphology in the backbone; the GSConv-based slim neck reduces part of the feature fusion cost; and an MSDA module in the high-resolution P3 branch enhances the context of small lesions. Following model selection on the validation split, the final locked models were evaluated once on the held-out test split, with HGSM-YOLO reaching 77.5% precision, 66.8% recall, 71.7% F1-score, 70.8% mAP@0.5, and 44.2% mAP@0.5:0.95, compared with 68.5%, 61.5%, 64.8%, 66.0%, and 40.2% for YOLO11n. A stratified outer five-fold cross-validation further yields 71.0% ± 1.4% mAP@0.5 and 44.4% ± 1.1% mAP@0.5:0.95 for HGSM-YOLO, versus 65.9% ± 1.1% and 40.2% ± 0.9% for YOLO11n. On the independent 1871-image citrus-leaf-disease-2 dataset, retraining under the same protocol gives 94.4% mAP@0.5 for HGSM-YOLO versus 92.2% for YOLO11n and 93.1% for the public Roboflow YOLOv11 reference model. The complete HGSM-YOLO architecture uses 7.2 GFLOPs, 2.82 M parameters, and runs at 90.9 FPS on the RTX 4090, compared with 6.4 GFLOPs, 2.59 M parameters, and 110.1 FPS for the baseline. Thus, the contribution provides a recall- and localization-oriented accuracy–efficiency trade-off rather than universal superiority in every individual metric. Full article

Continuous, non-invasive fish monitoring supports aquatic animal management, biodiversity assessment, and sustainable aquaculture, but embedded deployment requires a careful balance among accuracy, speed, memory, and computation under visually degraded underwater conditions. We developed ULFD-YOLO, an ultra-lightweight detector derived from YOLOv11n through coordinated redesign of the backbone, neck, and detection head. The model combines a custom convolutional MobileNetV4-tiny backbone, a hypergraph-based multi-scale fusion neck, and a lightweight MBConv head with channel attention. Experiments were conducted on Fish-BJ, an in-house dataset of 3402 images covering 21 species-informed aquarium-fish detection categories, and on a deliberately difficult 1180-image WildFish subset after dataset-specific training. On Fish-BJ, ULFD-YOLO achieved 0.960 mAP@0.5 and 0.732 mAP@0.5:0.95 with 1.3 M parameters, 2.6 GFLOPs, and a 3.0 MB model file, reducing parameters and computation by 50.0% and 58.7% relative to YOLOv11n. Bootstrap resampling yielded 95% confidence intervals of 0.946–0.973 and 0.638–0.821 for the two metrics, respectively. The model achieved 0.803 mAP@0.5 on WildFish and 19–24 FPS at 448 × 640 on a Jetson Orin Nano under its 15 W nvpmodel power mode. These results establish a practical accuracy–efficiency trade-off for embedded fish monitoring rather than peak localization accuracy. Full article

Image-level monitoring and variety-level detection of three specialty tomato cultivars, Kiss, Millennium, and White Jade, remain challenging in unstructured orchards because of foliage occlusion, overlapping fruit clusters, and variable illumination. Conventional downsampling may weaken fine spatial details of small targets, whereas larger detectors can impose computational demands that are unsuitable for mobile or edge-based agricultural platforms. To address these limitations, we propose WCAF-YOLO, a lightweight two-dimensional tomato detector based on a modified YOLOv26n architecture. The model replaces the P3 backbone downsampling operation with space-to-depth convolution (SPD-Conv) to retain fine-grained spatial information. Its weighted channel-aware fusion (WCAF) neck combines learnable branch weighting with parameter-free three-dimensional attention to refine fused features. Bounding-box regression uses focaler-minimum point distance intersection over union (Focaler-MPDIoU). Across five random seed runs on the internal held-out test subset of a custom single-site orchard dataset, WCAF-YOLO obtained a mean ${\mathrm{mAP}}_{50–95}$ of $0.9048\pm 0.0013$ and a mean recall of $0.9280\pm 0.0019$. The corresponding mean improvements over the YOLOv26n baseline were 2.14 and 3.42 percentage points, respectively. The model contained 2.36 M parameters and required 6.36 GFLOPs. Under the evaluated protocol, the model combined a compact parameter count with higher mean detection metrics than the YOLOv26n baseline. The detector outputs two-dimensional bounding boxes and variety labels for image-level orchard monitoring and variety-level assessment. Integration into agricultural field platforms remains to be validated. Full article

Efficient parking space management in urban settings represents a growing challenge owing to the sustained increase in the vehicle fleet. This study presents a comparative evaluation of five object detection architectures —YOLOv8s, YOLOv11s, YOLOv12s, RT-DETR-L and Faster R-CNN—applied to real-time intelligent vehicle occupancy monitoring under variable lighting conditions. The models were trained via transfer learning on a custom dataset of 1463 source images (21,944 annotated instances; expanded to 3511 files and 52,664 instances through offline augmentation of the training subset; three classes: free, occupied and unavailable) captured on a university campus located in Juliaca (Puno region), Peru, at 3824 m a.s.l. under daytime and nighttime clear-sky conditions from a single fixed-camera viewpoint. Each architecture was evaluated in ten independent experiments. Six dataset partitioning schemes of increasing strictness—a random control (R0) plus five leakage-controlled partitions—were evaluated. Under the strictest scheme (D3), simultaneously disjoint in acquisition date and camera viewpoint and therefore the most rigorous generalization estimate obtained in this study, accuracy ranges from mAP@0.5:0.95 of 0.9325 for Faster R-CNN to 0.8763 for YOLOv11s. Under the random partitioning conventionally applied to fixed-camera datasets, the same five architectures fell within 0.0055 of one another, all above 0.985, and their ranking was essentially inverted (Spearman $\rho =-0.80$). The differences in computational efficiency across architectures were statistically significant ($H=47.06$, $p<0.001$). YOLOv8s was the fastest of the four non-dominated architectures under the disjoint partition and was selected in 73.3% of weightings, although it ranked fourth in accuracy; its recommendation therefore rests on computational efficiency under a real-time constraint, whereas deployments that prioritize accuracy are better served by Faster R-CNN. The integrated system YOLOv8s + ByteTrack + FastAPI + Next.js 14 achieved per-slot accuracies of 87.5% and 91.8% under daytime and nighttime clear-sky conditions, respectively, using 1395 observations collected in a single university parking lot. For YOLOv8s, the transition from random to disjoint partitioning costs 0.1085 in mAP@0.5:0.95 (0.9913 to 0.8828), indicating that the near-saturated performance obtained under random partitioning substantially reflects the memorization of a fixed spatial configuration rather than generalization. The results support the feasibility of single-stage CNN architectures for intelligent parking monitoring in high-altitude Andean university environments under the evaluated acquisition conditions. Full article

A lightweight multi-scale group attention and channel dynamic residual fusion method is proposed to address the challenge of detecting extremely small encoded targets in particle accelerator multipole magnet collimation. In this scenario, the encoded targets occupy merely 0.003% to 0.009% of the total image pixels, rendering conventional visual techniques inadequate for capturing such fine-grained features. The paper proposes MLGA-CDRF-YOLO, an improved detection framework based on YOLOv11s that integrates a lightweight multi-scale group attention mechanism with channel-wise dynamic residual fusion. Specifically, the C2f-CDRF module combines lightweight channel attention with a dynamic residual structure, enhancing feature representation while reducing parameter count and computational complexity. The Multi-scale Lightweight Group Attention (MLGA) module splits input features into groups processed in parallel through a channel attention branch and a multi-scale spatial attention branch, with ChannelShuffle enabling cross-group information exchange, to improve the model’s feature representation capability and generalization performance for complex scenes. Experimental results on the encoded target dataset demonstrate that MLGA-CDRF-YOLO achieves an mAP@0.5 of 95.5%, a precision of 95.8%, and a recall of 88.5%, with only 3.41 M parameters and 14.8 GFLOPs, achieving a competitive and comprehensive balance between accuracy and computational cost. Furthermore, evaluation on the NEU-DET dataset confirms the model’s stable generalization performance across diverse detection tasks. Full article

UAV-based insulator detection is challenging because insulators often occupy small regions of aerial images and appear against complex backgrounds, while subtle local features may be lost during feature extraction and down-sampling. Moreover, practical UAV and edge-device applications require efficient models with limited computational and memory demands. This study proposes an optimized lightweight YOLOv11n model for high-voltage transmission-line insulator detection. The architecture integrates C3k2MBNV2 to reduce model complexity, SCDown to preserve spatial information during down-sampling, and C3k2WTDC to enhance multi-frequency feature representation. A diverse dataset containing 5750 insulator images acquired under different environmental conditions, viewing angles, and backgrounds was used for evaluation. Experimental results show that the proposed model reduces the parameter count from 6.20 M to 3.26 M and computational complexity from 20.5 to 12.7 GFLOPs, corresponding to reductions of 47.4% and 38.0%, respectively. Meanwhile, precision increases from 91.3% to 93.8%, recall from 73.4% to 75.2%, mAP50 from 71.2% to 73.9%, and mAP50–95 from 65.6% to 67.3%. These results demonstrate an improved accuracy–efficiency trade-off, supporting real-time insulator detection in resource-constrained UAV and edge-device applications. Full article

Power line component and insulator defect detection is an important task in intelligent transmission line inspection. However, UAV-based inspection images often contain small, densely distributed targets under complex backgrounds, making it difficult for existing detectors to balance accuracy and model compactness. To address these challenges, this paper proposes a lightweight YOLOv8-based detector named DSCMamba-TAD-YOLOv8. First, depthwise separable convolutions are introduced into the Neck to reduce parameters and computational cost. Second, DSCMambaNet replaces the original C2f module to enhance multi-scale feature representation by combining lightweight local feature extraction and cross-region contextual modeling. An embedded CBAM component is further integrated inside DSCMambaNet to strengthen informative channel responses and spatial regions. Finally, a Task-Aware Dynamic Detection Head, named TADetect, improves head adaptability through scale-aware and task-aware feature modulation. Experiments on the InsPLAD-det dataset show that DSCMamba-TAD-YOLOv8 achieves 91.86% Precision, 88.02% Recall, 91.83% mAP@0.5, and 74.82% mAP@0.5:0.95. Compared with YOLOv8n, the proposed model improves Precision, mAP@0.5, and mAP@0.5:0.95 by 4.09, 2.43, and 4.46 percentage points, respectively, while maintaining a comparable Recall level with a slight increase from 87.04% to 88.02%. Meanwhile, Params decrease from 3.209 M to 2.702 M and GFLOPs from 8.2 to 7.5. On the revised TPL-SOD held-out test subset, the proposed model improves Precision from 86.20% to 88.16%, mAP@0.5 from 87.09% to 88.81%, and mAP@0.5:0.95 from 68.44% to 70.13%, while Recall remains stable and slightly increases from 91.75% to 92.33%. These results demonstrate that DSCMamba-TAD-YOLOv8 improves detection accuracy and localization quality while maintaining a compact structure and stable recall performance. Full article

While deep learning has significantly improved the accuracy of visual detection systems, the integration of process semantics with visual perception for industrial assembly tasks remains largely unexplored. This study aims to develop a high-accuracy cable connector detection framework that incorporates process knowledge to enhance feature discrimination under varying industrial imaging conditions. To achieve this goal, we build an enhanced Fully Convolutional One-Stage (FCOS) detector with a ConvNeXt V2 backbone and introduce a Process Feature Linear Modulation (PFNM) module. The proposed module adaptively modulates visual features using encoded process semantics, enabling the detector to align visual perception with assembly logic. Experiments conducted on an industrial connector dataset demonstrate that the proposed method achieves an mAP of 84.7%, outperforming representative state-of-the-art detectors including YOLOv11, RT-DETR, and DINO while maintaining an inference speed of 17.6 FPS. Ablation studies further show that each component contributes to progressive performance improvement, and the complete framework achieves a 5.5% AP gain over the ResNet-50 baseline. These results indicate that integrating process knowledge with visual feature learning effectively improves feature discrimination and provides a promising paradigm for process-aware perception in intelligent manufacturing. Full article

Objective: To address small bolt defect targets, complex background interference, and limited edge deployment in hilly and mountainous rail transport environments, a detection method balancing accuracy, lightweight design, and real-time performance was proposed. Methods: A track image dataset containing missing bolts, loose bolts, and missing nuts was constructed. Based on YOLOv11m, HFERBC3K2 was developed by replacing the standard bottleneck in C3K2 with a High-Frequency Enhancement Residual Block to strengthen edge, texture, and local structural feature extraction. A Spectral Enhanced Feed-Forward module was introduced into C2PSA to form SEFFNC2PSA, enhancing defect-related frequency components and suppressing background interference through adaptive frequency-domain modulation. The integrated model was named HFSE-YOLOv11. Channel-level structured pruning was then applied, and the model with a pruning ratio of 0.5 was named LHFSE-YOLOv11. Results: On the validation set, HFSE-YOLOv11 achieved 91.7% precision, 93.4% recall, 91.3% mAP@0.5, and 80.2% mAP@0.5:0.95, improving upon YOLOv11m by 3.6, 2.2, 1.2, and 3.1 percentage points, respectively. After pruning, LHFSE-YOLOv11 had 15.9 M parameters, 53.8 GFLOPs, and a 32.5 MB model size, representing reductions of 16.3%, 14.3%, and 11.7%, while mAP@0.5 and mAP@0.5:0.95 decreased by only 0.3 and 0.9 percentage points. On the independent test set, it achieved 91.7% precision, 93.4% recall, 91.0% mAP@0.5, 79.3% mAP@0.5:0.95, and 81.5 FPS, outperforming all compared models in the four detection metrics. Conclusion: LHFSE-YOLOv11 balances accuracy, efficiency, and model size, supporting deployment on vehicle-mounted inspection terminals and resource-constrained edge devices. Full article

Quality control in green coffee bean production is critical for food safety and economic sustainability. A persistent gap in existing automated inspection systems is their inability to capture both sides of each bean, which leads to systematic under-detection of surface defects. This work presents three contributions: (i) a novel mechatronic apparatus that mechanically guarantees dual-sided imaging of every bean, (ii) a public 12-class dataset of green coffee bean defects, and (iii) an embedded, real-time inspection pipeline validated on low-cost hardware. The apparatus sequentially presents each bean, from a standard 350 g sample, to two 16-megapixel cameras under controlled LED illumination. A dataset of 9600 images spanning 12 classes (11 defects and 1 normal) was generated from expert-classified samples and enriched through data augmentation. Four convolutional neural network (CNN) architectures, VGG-16, VGG-19, ResNet-50 and YOLOv8, were trained and benchmarked using precision, recall, F1-score and mean average precision. YOLOv8 achieved the best overall performance, with a precision of 97.4%, a recall of 99.6%, an F1-score of 0.930 and a mean average precision of 96.5%, outperforming VGG-16 (accuracy 86.07%), VGG-19 (accuracy 67.03%) and ResNet-50 (accuracy 87.76%). Dual-sided acquisition raised mean per-class detection accuracy from 0.727 to 0.908, a relative gain of 25.7% over an equivalent single-sided configuration. Deployed in real-time “track” mode on a Raspberry Pi 4, the system simultaneously classifies defects and counts beans by category, processing a 350 g sample in approximately 38 min. Combining mechanical innovation with lightweight deep learning enables practical, scalable, and cost-effective quality control for laboratories specialized in coffee analysis. Full article

Accurate shank-length phenotyping of live chickens is hindered by feather occlusion and uncertain endpoint localization in visible images. We developed a two-stage method that fuses registered visible and infrared images before predicting the two measurement endpoints. The fusion network used a U-Net encoder–decoder with residual blocks, Coordinate Attention, and Strip Pooling, trained with a YUV-guided loss. A YOLOv8s-Pose model with Coordinate Attention and a length-related loss then localized the endpoints. The dataset comprised 100 chickens and 1000 paired visible–infrared acquisitions, separated at the individual level into training, validation, and test sets. On the test set, at the chicken level, the reported mean signed difference, mean absolute error, and root mean square error were 0.129 mm, 0.790 mm, and 0.984 mm, respectively. The Pearson correlation coefficient between model-derived and manual reference measurements was 0.992. Compared with either single-modality input, the fused images reduced the mean absolute error and root mean square error. These findings show that complementary texture and thermal-boundary information can support accurate vision-based shank-length measurement under controlled acquisition conditions. Full article

In protected agriculture such as greenhouses, pest and disease monitoring via UAVs and fixed cameras suffers from extremely small object proportions owing to shooting altitude constraints, posing considerable detection challenges. Moreover, fine-grained annotation of numerous small-scale images incurs prohibitive costs. Targeting this bottleneck, this paper proposes a cross-scale unified semantic space learning framework and introduces an end-to-end DS-DETR detector based on DETR. Unlike existing methods relying on domain adaptation, multi-scale fusion, or super-resolution reconstruction, this work explicitly models instance-level cross-scale semantic correlation, transferring fine-grained semantics from large-scale close-up images to small-scale scene feature space. A Single-Point Dual-Shooting (SPDS) strategy is adopted to collect high-fidelity paired images via ordinary smartphones at low cost. A dual-stream encoder with cross-view attention and an instance-level contrastive loss align features of identical instances in a unified semantic space. A self-built CropScale-Det dataset covering three crop diseases is constructed in greenhouse scenarios. Experimental results show that DS-DETR achieves 42.5 ± 1.2% mAP@50 under limited annotations, outperforming YOLOv8-n by 11.2%, with small-target average precision reaching 26.8 ± 1.1%. Ablation experiments and feature visualization validate the effectiveness of the designed mechanism. This approach considerably reduces reliance on large-scale densely annotated data, establishing a data-efficient proof-of-concept for small-scale pest detection in protected agriculture. Full article

Accurate pavement distress identification is essential for infrastructure maintenance. However, prevailing models frequently underperform in complicated environments due to extreme scale variations, atypical defect geometries, and severe background noise. To mitigate these limitations, this study presents BDC-YOLO, an upgraded detection network built upon the YOLOv8 baseline. The proposed architecture structurally incorporates three specialized mechanisms: Bi-level Routing Attention (BRA) to isolate relevant target features from background artifacts; Dynamic Snake Convolution (DySnakeConv) to capture the topological characteristics of elongated and irregularly shaped cracks; and Content-Aware ReAssembly of FEatures (CARAFE) to minimize information degradation during upsampling and refine multi-scale feature fusion. Evaluated on the RDDChina dataset, BDC-YOLO demonstrates superior accuracy over the baseline and comparative state-of-the-art methods. Specifically, the framework yields a mAP_0.5 of 88.9%, representing an absolute gain of 4.7% against the standard YOLOv8 model while achieving an inference speed of 175.4 FPS. Full article

Overcoming complex background noise and poor small-target detection in industrial settings, this paper introduces YOLOv8-BBP2, an enhanced YOLOv8 model. To better extract dynamic features, the backbone integrates a BiFormer dual-level routing attention mechanism. Moreover, a learnable Bi-directional Feature Pyramid Network (BiFPN) replaces the standard module, optimizing multi-scale feature integration. A P2 detection head is also added to accurately identify tiny objects, such as early flames and thin smoke. Tested on a custom factory fire dataset, YOLOv8-BBP2 yields 95.231% precision, 94.612% recall, and 89.677% mean average precision (mAP@0.5). These metrics represent respective gains of 3.31%, 4.934%, and 7.451% over the baseline YOLOv8s. Ultimately, with an inference speed of 20 ms per frame, the proposed network ensures highly robust, real-time performance. Full article

This paper focuses on the binary task of parking-slot occupancy detection (vacant vs. occupied) from UAV aerial imagery. Accurate parking-slot detection from UAV imagery is challenged by small target sizes, highly regular rectangular shapes, large-scale variations, complex backgrounds, and perspective distortions. Compared with fixed surveillance cameras, UAV-based detection offers flexible deployment, wide-area coverage, and no requirement for pre-installed infrastructure, making it especially suitable for large open-air parking lots and temporary parking scenarios. To address this, this study proposes a task-specific framework for UAV-based parking-slot detection, with improvements in backbone design, attention modeling, and bounding-box regression. A lightweight MnasNet-inspired backbone is used to improve multi-scale feature extraction at low computational cost. An enhanced EMA module with adaptive grouping, FFT-based frequency enhancement, and gated fusion is introduced to better model the structured patterns of parking lot scenes. In addition, a UIoU+ loss tailored to rectangular geometry is proposed to improve localization quality. Sensitivity analysis and repeated experiments show that the method is stable and statistically reliable. All main metrics are evaluated on an independent held-out test set to ensure generalization. Extensive experiments demonstrate that each component brings consistent performance gains. The proposed model achieves 99.44 ± 0.12% mAP@0.5, 90.31 ± 0.27% mAP@0.5:0.95, 99.27 ± 0.15% precision, and 99.00 ± 0.18% recall on the self-built UAV Parking Lot dataset. Its mAP@0.5:0.95 is 26.01 percentage points higher than the YOLOv11n baseline. Consistent performance improvements are also validated on two additional public benchmarks (CARPK and PKLot), confirming the generalization of the proposed method beyond the self-built dataset. Most importantly, the method supports real-time inference on embedded UAV platforms and achieves state-of-the-art performance among lightweight detectors, making it an ideal solution for practical intelligent parking management. Ablation studies further confirm the complementary synergy between the proposed backbone, attention module, and loss function. Full implementation code, pre-trained weights, and detailed reproduction guidelines are publicly available to ensure research reproducibility. Full article