TL;DR
OrchardDA-YOLO is introduced, a semi-supervised domain adaptation detection framework oriented toward edge deployment that combines a foreground-aware Task-aware Coordinate Attention module with a progressive cross-domain alignment strategy that enhances the model’s feature discriminability for occluded and small-scale targets.
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.
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.
In intelligent transportation systems, real-time detection demands high inference speed, making single-stage detectors the preferred choice. However, existing semi-supervised object detection (SSOD) frameworks suffer from pseudo-label noise and limited multi-scale adaptability when applied to such detectors. This paper presents AD-YOLO, a dual-level framework that tackles these issues from both the detector architecture and the SSOD pipeline. At the detector level, a normalization-guided attention module enables feature recalibration with zero extra parameters, and a CIoU-NWD hybrid loss incorporates the Wasserstein distance to suppress localization jitter. At the framework level, an Adaptive Teacher employs a category-aware dynamic threshold that selects pseudo-labels based on confidence percentiles, eliminating preset thresholds and calibration bias. This thresholding is assisted by a scale-aware dynamic augmentation mechanism, which uses teacher-generated pseudo-labels to identify small-object images and applies only weak augmentation to preserve their semantics. These two levels form a self-reinforcing loop. Experiments on BDD100K and TT100K show that with only 10% labeled data, AD-YOLO achieves 28.83% mAP at 142 FPS, demonstrating better performance compared with peer methods and exhibiting robustness in complex traffic scenes.
Pineapple detection in complex field environments faces significant challenges, including severe background interference, frequent occlusion by sword leaves, and difficulties in identifying dense fruits, which demand an optimal balance between detection accuracy and computational efficiency for practical deployment on resource-constrained edge devices. We propose YOLO-PINE, a lightweight detection model based on an improved YOLO11n architecture incorporating three key modules: the Channel Grouping semi-convolution module (CGHalfConv) for efficient shallow feature extraction, the Circular Attention mechanism (CA) for global context modeling in the frequency domain, and the Spatial Attention Multi-scale Convolution module (SAMC) for enhanced multi-scale feature fusion. The model was trained and evaluated on a self-constructed pineapple dataset. YOLO-PINE achieved a precision of 97.2%, a recall of 92.6%, mAP@50 of 95.9%, and mAP@50–95 of 70.1%, with only 5.0 GFLOPs and 2.1 M parameters, representing a 21.9% reduction in computational load and a 16.0% decrease in parameters compared to YOLO11n. YOLO-PINE achieves a competitive balance between detection accuracy and efficiency, offering a viable visual perception solution for automated pineapple harvesting on edge devices.
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 detection of small, rotated objects in practical scenarios such as remote sensing imagery and autonomous driving remains a critical challenge. Existing methods often struggle to balance detection accuracy and real-time performance, and face limitations in capturing subtle features, fusing multi-scale information, and regressing oriented bounding box angles. To address these issues, this paper proposes an Adaptive and Scalable YOLO model named AS-YOLOR (Adaptive and Scalable YOLO for Rotated object detection), based on the YOLOv8 baseline. The model introduces systematic innovations at three levels: feature enhancement, attention mechanism, and feature fusion. Firstly, a high-resolution detection head is introduced, combined with a Cross-layer Soft-weighted Feature Fusion (CSFF) strategy, to adaptively fuse shallow detail information with deep semantic features, thereby significantly improving the model’s ability to capture features of tiny objects. Secondly, a Hybrid Attention Module (HAM), enhanced with global context, is designed. This module integrates spatial attention in parallel with an improved Global Context Aware Block, effectively strengthening the model’s focus on and discriminative power for key target regions. Finally, an Adaptive Feature Fusion Path Network (AFFPN) is constructed. By incorporating learnable weights and scale-sequence modeling, it achieves superior modeling of the nonlinear relationships between multi-scale features. Experimental results on the DOTA-V1.0 public dataset demonstrate that the proposed AS-YOLOR model achieves 82.72% on the core evaluation metric mAP@50, representing a significant improvement of 3.22 percentage points over the baseline YOLOv8 model, and outperforming various mainstream rotated object detection methods. Comprehensive ablation studies validate the effectiveness of each improved component within the model. This research provides a solution with strong practical potential for achieving efficient and high-precision detection of small, rotated objects.