The real-time detection of banana bunches and stalks in banana orchards is a key technology in the application of agricultural robots. The complex conditions of the orchard make accurate detection a difficult task, and the light weight of the deep learning network is an application trend. This study proposes and compares two improved YOLOv4 neural network detection models in a banana orchard. One is the YOLO-Banana detection model, which analyzes banana characteristics and network structure to prune the less important network layers; the other is the YOLO-Banana-l4 detection model, which, by adding a YOLO head layer to the pruned network structure, explores the impact of a four-scale prediction structure on the pruning network. The results show that YOLO-Banana and YOLO-Banana-l4 could reduce the network weight and shorten the detection time compared with YOLOv4. Furthermore, YOLO-Banana detection model has the best performance, with good detection accuracy for banana bunches and stalks in the natural environment. The average precision (AP) values of the YOLO-Banana detection model on banana bunches and stalks are 98.4% and 85.98%, and the mean average precision (mAP) of the detection model is 92.19%. The model weight is reduced from 244 to 137 MB, and the detection time is shortened from 44.96 to 35.33 ms. In short, the network is lightweight and has good real-time performance and application prospects in intelligent management and automatic harvesting in the banana orchard.
Intelligent Phytoprotection is an important direction for the modern development of plant protection related disciplines, and its essence is the innovative application of new generation information technology industry, high-end equipment manufacturing industry, and digital industry related technologies in the traditional plant protection field. This article first identifies 224 International Patent Classification (IPC) Main groups in the field of intelligent phytoprotection technology based on the International Patent Classification System. And then combines with China’s industrial policy practice, we explore the impact of industrial policy on the application number of invention patents in the field of intelligent phytoprotection technology using the Difference-in-difference (DID) method and the Synthetic DID method. The study results showed that the implementation of industrial policy can significantly promote the patent application activities in the intelligent phytoprotection treatment group, with an average increase of 517 invention patent applications compared to the control group that is not affected by the policy. The research conclusion of this article suggests that for countries and regions, industrial policies are an important tool for promoting the innovation and development of intelligent phytoprotection related technologies.
Considering the complex terrain of gobi orchards, including irrigation-induced orchard ditches and unstructured environments caused by low-hanging branches of spindle-shaped fruit trees, an improved A*-based regional path planning method is pro-posed for gobi orchards. First, penalty terms based on the distances to the orchard ditch centerline and edge were introduced during path searching. These terms provide adaptive constraints that guide the planned path toward the central region between fruit tree rows, effectively preventing the path from approaching orchard ditches and low-hanging branches. Second, to address the large spatial extent and high computational complexity of Gobi orchard environments, a semicircular arc search strategy was incorporated into the improved A* algorithm to enhance computational efficiency.Finally, the initial path was smoothed using a minimum turning radius circle and a dynamically adjusted two-line tangent strategy. This process generates a globally feasible path with continuous tangency, no sharp turning angles, compliance with the minimum turning radius constraint, and safe executability.The MATLAB simulation results showed that, compared with the conventional A* algorithm, the improved A* algorithm reduced the planning time and the number of searched nodes by 19.43% and 49.1%, respectively, while increasing the path length by 16.9%. Field test results showed that, although the overall path length increased, the improved A* algorithm exhibited high stability across repeated experiments and generated smoother paths. Therefore, the proposed method improves planning efficiency while generating safe and reliable paths under complex road conditions in gobi orchard environments.