This study conducted an analysis of zero-shot detection capabilities using two frameworks, YOLO-World and Grounding DINO, on a selection of images in the wild blueberry (Vaccinium angustifolium Ait.) cropping system. The datasets included ripe wild blueberries, hair fescue (Festuca filiformis Pourr.), blueberry buds, and red leaf disease (Exobasidium vaccinii). Key performance metrics such as Intersection over Union (IoU), precision, recall, and F1 score were utilized for model comparison. Grounding DINO consistently achieved superior performance across all metrics and datasets, achieving significantly higher mean IoUs on berries, red leaf, hair fescue, and buds (0.642, 0.921, 0.735, and 0.629, respectively) compared to YOLO-World (0.516, 0.567, 0.232, and 0.408, respectively). Evidenced by their high recall rates relative to precision, the models displayed a preference for identifying true positives at the cost of increasing false positives. Grounding DINO’s higher precision (overall mean of 0.672), despite the tendency to over-detect, indicated a better balance in minimizing false positives than YOLO-World (overall mean of 0.501). These findings contrast with the foundational study of YOLO-World where it demonstrated superior performance on standard datasets, highlighting the importance of dataset characteristics and optimization processes in model performance. The practical implications of this study include providing a solution for accelerated object detection image annotation in the wild blueberry cropping system. This work, representing a significant advancement in facilitating accurate and efficient annotation of wild blueberry datasets, guides future research in the application of zero-shot detection models to agricultural datasets.

This research addresses the critical need for efficient image annotation in precision agriculture, using the wild blueberry (Vaccinium angustifolium Ait.) cropping system as a representative application to enable data-driven crop management. Tasks such as automated berry ripeness detection, plant disease identification, plant growth stage monitoring, and weed detection rely on extensive annotated datasets. However, manual annotation is labor-intensive, time-consuming, and impractical for large-scale agricultural systems. To address this challenge, this study evaluates an automated annotation pipeline that integrates zero-shot detection models from two frameworks (Grounding DINO and YOLO-World) with the Segment Anything Model version 2 (SAM2). The models were tested on detecting and segmenting ripe wild blueberries, developmental wild blueberry buds, hair fescue (Festuca filiformis Pourr.), and red leaf disease (Exobasidium vaccinii). Grounding DINO consistently outperformed YOLO-World, with its Swin-T achieving mean Intersection over Union (mIoU) scores of 0.694 ± 0.175 for fescue grass and 0.905 ± 0.114 for red leaf disease when paired with SAM2-Large. For ripe wild blueberry detection, Swin-B with SAM2-Small achieved the highest performance (mIoU of 0.738 ± 0.189). Whereas for wild blueberry buds, Swin-B with SAM2-Large yielded the highest performance (0.751 ± 0.154). Processing times were also evaluated, with SAM2-Tiny, Small, and Base demonstrating the shortest durations when paired with Swin-T (0.30–0.33 s) and Swin-B (0.35–0.38 s). SAM2-Large, despite higher segmentation accuracy, had significantly longer processing times (significance level α = 0.05), making it less practical for real-time applications. This research offers a scalable solution for rapid, accurate annotation of agricultural images, improving targeted crop management. Future research should optimize these models for different cropping systems, such as orchard-based agriculture, row crops, and greenhouse farming, and expand their application to diverse crops to validate their generalizability.

This study introduced a novel approach to 3D image segmentation utilizing a neural network framework applied to 2D depth map imagery, with Z axis values visualized through color gradation. This research involved comprehensive data collection from mechanically harvested wild blueberries to populate 3D and red–green–blue (RGB) images of filled totes through time-of-flight and RGB cameras, respectively. Advanced neural network models from the YOLOv8 and Detectron2 frameworks were assessed for their segmentation capabilities. Notably, the YOLOv8 models, particularly YOLOv8n-seg, demonstrated superior processing efficiency, with an average time of 18.10 ms, significantly faster than the Detectron2 models, which exceeded 57 ms, while maintaining high performance with a mean intersection over union (IoU) of 0.944 and a Matthew’s correlation coefficient (MCC) of 0.957. A qualitative comparison of segmentation masks indicated that the YOLO models produced smoother and more accurate object boundaries, whereas Detectron2 showed jagged edges and under-segmentation. Statistical analyses, including ANOVA and Tukey’s HSD test (α = 0.05), confirmed the superior segmentation performance of models on depth maps over RGB images (p < 0.001). This study concludes by recommending the YOLOv8n-seg model for real-time 3D segmentation in precision agriculture, providing insights that can enhance volume estimation, yield prediction, and resource management practices.