In the context of blueberry orchard management and automated harvesting, this study introduces an improved YOLOv8 model, ADE-YOLO, designed for precise blueberry ripeness detection, enhancing automated picking efficiency. Built on the YOLOv8n architecture, ADE-YOLO features a dimensionality-reducing convolution at the backbone’s end, reducing computational complexity while optimizing input features. This improvement enhances the effectiveness of the AIFI module, particularly in multi-scale feature fusion, boosting detection accuracy and robustness. Additionally, the neck integrates a dynamic sampling technique, replacing traditional upsampling methods, allowing for more precise feature integration during feature transfer from P5 to P4 and P4 to P3. To further enhance computational efficiency, CIOU is replaced with EIOU, simplifying the aspect ratio penalty term while maintaining high accuracy in bounding box overlap and centroid distance calculations. Experimental results demonstrate ADE-YOLO’s strong performance in blueberry ripeness detection, achieving a precision of 96.49%, recall of 95.38%, and mAP scores of 97.56% (mAP50) and 79.25% (mAP50-95). The model is lightweight, with just 2.95 M parameters and a 6.2 MB weight file, outpacing YOLOv8n in these areas. ADE-YOLO’s design and performance underscore its significant application potential in blueberry orchard management, providing valuable support for precision agriculture.

The precise identification of maize kernel varieties is essential for germplasm resource management, genetic diversity conservation, and the optimization of agricultural production. To address the need for rapid and non-destructive variety identification, this study developed a novel interpretable machine learning approach that integrates low-field nuclear magnetic resonance (LF-NMR) with morphological image features through an optimized support vector machine (SVM) framework. First, LF-NMR signals were obtained from eleven maize kernel varieties, and ten key features were extracted from the transverse relaxation decay curves. Meanwhile, five image morphological features were selected using the recursive feature elimination (RFE) algorithm. Before modeling, principal component analysis (PCA) was used to determine the distribution features of the internal components for each maize variety. Subsequently, LF-NMR features and image morphological data were integrated to construct a classification model and the SVM hyperparameters were optimized using an improved differential evolution algorithm, achieving a final classification accuracy of 96.36%, which demonstrated strong robustness and precision. The model’s interpretability was further enhanced using Shapley values, which revealed the contributions of key features such as Max Signal and Signal at Max Curvature to classification decisions. This study provides an innovative technical solution for the efficient identification of maize varieties, supports the refined management of germplasm resources, and lays a foundation for genetic improvement and agricultural applications.