Aug 2026· Agriculture· Vol 16, pp. 1770· 0 citations· 25 references
Abstract
In complex and unstructured orchard environments, accurate fruit detection is essential for yield estimation and robotic harvesting in precision agriculture. However, single-model detectors often suffer from reduced robustness and high miss rates under drastic illumination changes, severe occlusions, and dense fruit overlap. To address these challenges, this study proposes a multi-model fusion framework for robust mango detection in complex orchard environments. The proposed method employs YOLOv8n, YOLOv8s, and YOLOv8m as base detectors and applies multi-scale test-time augmentation (TTA) to obtain predictions from different augmented views. After mapping the predicted bounding boxes back to the original image coordinate system, predictions corresponding to the same target across different TTA views of each base detector are matched based on the intersection over union (IoU), yielding model-specific prediction results. Weighted Box Fusion (WBF) is then applied to determine the fused bounding-box coordinates. For candidate targets jointly detected by multiple base detectors, the confidence scores provided by the individual models are combined using Noisy-OR to obtain the fused confidence score. Finally, Gaussian Soft-NMS is applied to decay the scores of overlapping candidate boxes, thereby reducing the risk of incorrectly suppressing adjacent mangoes in densely clustered scenes. Experiments on two complementary datasets under within-dataset evaluation protocols demonstrate the effectiveness of the proposed method. On the standard dataset (Data1), Recall and mAP@0.5 reach 95.52% and 98.60%, respectively. Across five repeated random holdout splits of Data2, the proposed framework increased the mean Recall from 82.79% to 84.90% and the mean mAP@0.5 from 90.27% to 91.23%. These results indicate that the proposed framework improves detection robustness and completeness compared with single-model detectors in complex orchard environments, demonstrating its potential for offline yield estimation and orchard phenotyping.
Accurate and real-time apple detection in complex orchard environments is essential for robotic harvesting but remains challenging because of illumination variation, foliage occlusion, and limited computational resources. This study proposes Fast-YOLO11n, a lightweight detector derived from the nano variant of You Only Look Once 11 (YOLO11n) and integrating three complementary components. A Fast-C3k2 module based on partial convolution (PConv) reduces redundant computation while preserving cross-layer feature transmission. A focal modulation (FM) mechanism enhances target-related responses and suppresses background interference under occlusion and uneven illumination. In addition, a parallel downsampling module, termed ADown, retains local geometric details and multi-scale semantic information during downsampling. Experiments were conducted on a field-collected orchard dataset comprising 2240 images and 22,673 annotated apple instances under diverse lighting, scale, and occlusion conditions. Fast-YOLO11n achieved mean average precision values of 75.76% across intersection-over-union (IoU) thresholds of 0.50–0.95 (mAP@50–95) and 91.29% at an IoU threshold of 0.50 (mAP@50), while operating at 366.19 frames per second (FPS) with 2.51 million parameters and 6.00 billion floating-point operations (FLOPs). Compared with the YOLO11n baseline, it improved mAP@50–95 and mAP@50 by 2.39 and 1.39 percentage points, respectively, while reducing the parameter count and FLOPs by 2.71% and 5.36%. Ablation experiments demonstrated the individual and combined effects of the three modules on detection performance and computational efficiency. The proposed model provides a favorable balance between detection accuracy and computational efficiency, indicating its potential for real-time orchard perception on resource-constrained platforms.
Jinan Gu, Zhong-Kai Shen, Juan Liu et al.· Agriculture· 0 citations
Agricultural pest detection is constrained by three critical challenges: small pest size causing feature loss, complex backgrounds (
e.g
., foliage and lighting) hindering feature extraction, and severe class imbalance (pests occupying <5% of image pixels), limiting standard detectors like You Only Look Once version 8 (YOLOv8). To overcome these, we propose Enhance-YOLOv8, which replaces YOLOv8's Cross Stage Partial with 2 convolutions (C2f) backbone module with Enhance Adaptive Fine-grained Channel Attention (Enhance_AFCA). This module integrates hierarchical multi-scale feature extraction and adaptive edge enhancement to address edge information loss and inadequate feature fusion. Its adaptive fine-grained channel attention dynamically weights informative channels while suppressing noise, reducing bias from class imbalance. A novel MANet_Pest Detection (MANet_PD) module in the neck network enhances multi-scale feature fusion, improving small-target detection. The model further adopts Wise Intersection over Union (WiseIoU) Loss with Inner-Focaler-ShapeIoU, accelerating convergence and refining localization
via
auxiliary bounding boxes and dynamic penalties. Experimental results demonstrate that Enhance-YOLOv8 achieves a mAP@0.5 of 93.5% on our custom dataset and 77.3% on the public AgroPest-12 dataset, representing improvements of 2.8 percentage points and 3.0 percentage points over YOLOv8, respectively. The model exhibits robust performance in complex agricultural scenarios and demonstrates strong potential for deployment in precision agriculture.
Saiqi Pi, Fa-Yuan Xu, Fei Wang et al.· PeerJ Computer Science· 0 citations
Accurate and robust object detection in complex disaster scenes is essential for effective emergency response; however, severe occlusion, dense overlap, and cluttered backgrounds pose significant challenges to conventional single-model detectors. To address these limitations, this study proposes a novel rescue-oriented detection framework that integrates a fine-grained disaster dataset, a cross-generational YOLO ensemble, and a consensus-based fusion strategy using Weighted Boxes Fusion (WBF). A dataset of 2323 images was constructed by re-annotating CDNIC19k with instance-level labels for four rescue-critical roles, enabling more precise evaluation in real-world scenarios. Heterogeneous YOLO models spanning multiple architectural generations were jointly exploited within a unified ensemble framework to leverage complementary representations. Meanwhile, a consensus-driven fusion strategy based on WBF was adopted to improve prediction aggregation in dense and occluded scenes. Experimental results showed that the proposed method outperformed single-model baselines and NMS-based approaches, improving mAP@0.5 from 0.696 to 0.756 (+6.0%) while maintaining strong recall and robustness. Analysis of the YOLOv12 family reveals an accuracy–efficiency trade-off, where lightweight models enable real-time inference while high-capacity models provide more reliable detection. Overall, these findings demonstrate that cross-generational architectural diversity combined with consensus-based fusion constitutes a generalizable and effective paradigm for high-precision disaster scene understanding under diverse deployment constraints.
Maize tassel detection is essential for maintaining genetic purity in hybrid seed production. However, existing methods often show limited performance in detecting small or occluded missed tassels and are difficult to deploy efficiently on resource-constrained field devices. To address these challenges, this study proposes MTDP-YOLO, a lightweight detection framework based on YOLOv11 for UAV-based missed tassel detection. The proposed framework integrates an HGNet backbone for efficient feature extraction, a BiFPN-GLSA dual-path feature fusion structure combined with FEFM to enhance multi-scale representation and small-target discrimination, and a layer-adaptive pruning strategy with channel-wise knowledge distillation to further compress the model while preserving accuracy. In addition, a dedicated in-domain UAV dataset containing 9,146 images after training-set augmentation was constructed to represent diverse field conditions, including complex backgrounds, illumination variation, and target occlusion. Experimental results show that MTDP-YOLO achieves 89.5% mAP@0.5 with only 1.6 M parameters and 5.8 GFLOPs, representing a 1.1-percentage-point improvement over YOLOv11n while reducing computational cost. After pruning and distillation, the compressed model maintains competitive detection performance and supports efficient offline inference on portable ground devices. Furthermore, a geographic coordinate extraction method was developed and integrated into a custom Missed Tassel Detection and Geolocation Software, enabling automatic conversion of detection results into georeferenced target locations. Field validation provided initial evidence of sub-meter positioning feasibility under the evaluated conditions, indicating its potential for practical post-detasseling inspection and precision field management in hybrid maize seed production.
Tiny-object detection in UAV aerial imagery remains challenging due to extremely small object scales, dense distributions, and complex backgrounds. Existing methods often suffer from inefficient query modeling and inadequate multi-scale feature representation, particularly in high-resolution scenarios with substantial variations in target density. To address these challenges, this paper proposes AQF-Net, a unified detection framework built upon the D-FINE architecture. AQF-Net integrates three key components: a Fixed-Query Self-Attention (FQSA) mechanism for efficient global context modeling, a Large-Receptive-Field Enhancement (LREA) module for enhanced multi-scale feature fusion, and an adaptive query modeling strategy for density-aware query allocation. These components are tightly coupled to jointly optimize feature representation and query generation, enabling the model to better adapt to complex UAV scenarios. Extensive experiments are conducted on the CODrone, VisDrone2019, and a self-constructed photovoltaic defect dataset (PV-DV). The results demonstrate that AQF-Net consistently outperforms the D-FINE baseline and several state-of-the-art methods in both overall detection accuracy and tiny-object detection capability. Notably, AQF-Net achieves 33.4% AP and 55.0% AP50 on the VisDrone2019 validation set, while maintaining a favorable balance between accuracy and computational efficiency.
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 mAP50–95 of 0.9048±0.0013 and a mean recall of 0.9280±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.
Xudong Lin, Yi-Hao Zhang, Xianzhi Tu et al.· Horticulturae· 0 citations