Synchronized Detection of Tomato Plant Diseases, Pests, and Maturity on E-Commerce Platforms

针对电商场景下番茄品质管控与供应链响应效率的协同需求，本文提出一种面向电子商务的改进YOLOv8番茄植株病虫害与成熟度同步检测方法。通过深度学习模型，实现植株生长状态的多维度感知。技术改进方面，将原本的Conv模块替换为AKConv模块，并添加SimAM注意力机制，引入了MPDIoU损失函数去避免引入复杂的惩罚项降低计算复杂度，使模型计算量降低16.35%的同时维持82.4%的检测精度。实验表明，在自建包含6类常见病害与3级成熟度的番茄数据集上，模型综合检测精度达82.4%。In response to the collaborative demand for tomato quality control and supply chain response efficiency in e-commerce scenarios, this paper proposes an improved YOLOv8 tomato plant disease and pest synchronous detection method for e-commerce. Realize multi-dimensional perception of plant growth status through deep learning models. In terms of technological improvements, the original Conv module was replaced with the AKConv module, and SimAM attention mechanism was added. The MPDIoU loss function was introduced to avoid introducing complex penalty terms and reduce computational complexity, resulting in a 16.35% reduction in model computation while maintaining 82.4% detection accuracy. The experiment showed that the comprehensive detection accuracy of the model reached 82.4% on a self-built tomato dataset containing 6 common diseases and 3 levels of maturity.