Xiaopeng Yang

Fruit occlusion, dense clustering, platform vibration, and cross-frame re-identification errors reduce the reliability of video-stream yield estimation in high-density dwarf apple orchards. To address these limitations, this study developed a vision-based framework for apple detection, tracking, dynamic counting, and yield estimation. Apple detection was improved by embedding a Shuffle Attention (SA) module into YOLOv12n and introducing SlideLoss for hard-sample reweighting, yielding the YOLOv12n-SA-SlideLoss detector. Following detection, the DeepSORT tracking module was also improved by augmenting it with global motion compensation (GMC) and a dual-criteria association strategy based on motion consistency and appearance similarity. Yield estimation was formulated as a counting-based compensation model using dynamic fruit counts, mean fruit mass, and an occlusion compensation factor, and was compared with a pixel-geometric regression model based on detection boxes. Field experiments were conducted in a high-density dwarf ‘Fuji’ apple orchard in Aksu, Xinjiang, China. YOLOv12n-SA-SlideLoss achieved an mAP@0.5 of 95.7%, exceeding the YOLOv12n baseline by 2.3%. The improved DeepSORT reached 87.4% multiple object tracking accuracy and 86.7% multiple object tracking precision, while reducing identification switches by 22.2% relative to the original algorithm. The dynamic counting framework increased average counting accuracy from 85.6% to 93.5%. For yield estimation, the counting-based compensation model achieved a mean prediction accuracy of 85.2%, an RMSE of 18.5 kg, and a mean bias of −17.2 kg across six validation samples (n = 6), each covering multiple trees. Overall, the results indicate that combining detection, motion-compensated tracking, and counting-based correction improves stability of single-view video-stream yield estimation in high-density dwarf apple orchards.

Efficient vibration based walnut harvesting relies on the accurate determination of clamping-points on tree trunks. Properly selected clamping points can significantly enhance vibration transmission efficiency while minimizing mechanical damage to trees. However, most existing studies focus on developing generalized vibration models applicable to multiple trees, often overlooking the structural uniqueness of individual trees in clamping-point determination. This study proposes a monocular vision-based method for clamping-point determination in walnut vibration harvesting. Robustness and applicability in complex orchard environments are enhanced by introducing three keypoint annotation strategies with varying levels of structural constraints, namely 5-key-part annotation (5-KAS), 2-key-part annotation (2-KAS), and single-key-part annotation (1-KAS). Pose estimation models based on the YOLO architecture, including YOLOv8-pose, YOLO11-pose, and YOLOv12-pose, were evaluated to examine the effect of structure assisted annotation, and the results show that introducing structural constraints improves detection accuracy, training stability, and robustness. The YOLOv12-pose model combined with the 5-KAS achieves the best performance, with a precision of 95.8% and mean average precision (mAP) of 95.5%. Field harvesting experiments demonstrate that clamping-point prediction incorporating structural information achieves higher and more stable net harvesting rates. Overall, the proposed method offers a reliable and deployable solution for clamping-point determination using monocular RGB images, facilitating intelligent vibration harvesting in walnut orchards.