Jixin fruit (Malus domestica ‘Jixin’) is a high-value specialty fruit of significant economic importance in northeastern and northwestern China. Automatic recognition of fruit maturity is a critical prerequisite for intelligent harvesting. However, challenges inherent to field environments—including heterogeneous ripeness levels among fruits on the same plant, gradual color transitions during maturation that result in ambiguous boundaries, and occlusion by branches and foliage—render traditional image recognition methods inadequate for simultaneously achieving high recognition accuracy and computational efficiency. Although existing deep learning models can improve recognition accuracy, their substantial computational demands and high hardware requirements preclude deployment on resource-constrained embedded devices such as harvesting robots. To achieve the rapid and accurate identification of Jixin fruit maturity, this study proposes Multi-Attention DeepLabV3+ (MA-DeepLabV3+), a streamlined semantic segmentation framework derived from an enhanced DeepLabV3+ model. First, a lightweight backbone network is adopted to replace the original complex structure, substantially reducing computational burden. Second, a Multi-Scale Self-Attention Module (MSAM) is proposed to replace the traditional Atrous Spatial Pyramid Pooling (ASPP) structure, reducing network computational cost while enhancing the model’s perception capability for fruits of different scales. Finally, an Attention and Convolution Fusion Module (ACFM) is introduced in the decoding stage to significantly improve boundary segmentation accuracy and small target recognition ability. Experimental results on a self-constructed Jixin fruit dataset demonstrated that the proposed MA-DeepLabV3+ model achieves an mIoU of 86.13%, mPA of 91.29%, and F1 score of 90.05%, while reducing the number of parameters by 89.8% and computational cost by 55.3% compared to the original model. The inference speed increased from 41 frames per second (FPS) to 81 FPS, representing an approximately two-fold improvement. The model memory footprint is only 21 MB, demonstrating potential for deployment on embedded devices such as harvesting robots. Experimental results demonstrate that the proposed model achieves significant reductions in computational complexity while maintaining high segmentation accuracy, exhibiting robust performance particularly in complex scenarios involving color gradients, ambiguous boundaries, and occlusion. This study provides technical support for the development of intelligent Jixin fruit harvesting equipment and offers a valuable reference for the application of lightweight deep learning models in smart agriculture.

Potato foliar diseases pose a serious threat to crop growth and yield, making timely and accurate disease detection essential for early intervention and ef

Addressing technical challenges such as inaccurate appearance detection, inaccurate weight estimation, and low automation levels in passion fruit sorting under postharvest conveyor-line conditions, this study proposes an intelligent detection and grading model, YOLOv11n-ACH, based on an improved YOLOv11n. By integrating the Hybrid Inverted Block (HIB) module, the ASF-YOLO scale fusion mechanism, and the Convolutional Attention Fusion Mechanism (CAFM), the model effectively mitigates severe fruit occlusion and background interference caused by conveyor surfaces, residual plant material, and illumination variation. Consequently, the high-precision metric mAP@50-95 reaches 99.4%, representing an increase of 3.9 percentage points over the baseline model. Building upon this foundation, a real-time grading and counting system incorporating a confidence-priority frame-selection mechanism was constructed by combining the ByteTrack multi-object tracking algorithm with horizontal dynamic scale calibration technology. The study establishes a multivariate linear regression mass-estimation model based on morphological features (R² = 0.9617). The regression model was developed using 500 fruits, and its performance was independently evaluated using a second, non-overlapping cohort of 500 fruits collected from the same orchard. Using 12 horizontal calibration points, a cubic spline interpolation function was constructed to compensate for horizontal position-dependent variation in the pixel-to-physical scale under the tested fixed imaging configuration. In the independent mass-validation cohort, the system achieved an MAE of 2.54 g, an RMSE of 3.21 g, and an MARE of 5.65%. A third, non-overlapping cohort of 1568 fruits was used for end-to-end passage-level counting and operational grading evaluation. This lightweight solution provides an engineering approach for passion-fruit sorting under the tested postharvest conveyor-line conditions.