自定义苹果树数据集

本研究使用Stable Diffusion 2.1-base生成了一个自定义的苹果树数据集，用于训练YOLOv5m对象检测模型。该数据集包含670张通过提示工程生成的逼真苹果树图像，旨在模拟真实世界中的苹果检测场景。数据集的创建过程涉及使用预训练的Stable Diffusion模型，通过精心设计的正负提示生成图像，并进行自动标注以确定苹果的位置。此数据集的应用领域主要集中在无人机系统的苹果检测，以提高农业监测和基础设施检查的效率。

This study utilized Stable Diffusion 2.1-base to generate a custom apple tree dataset for training the YOLOv5m object detection model. The dataset includes 670 photorealistic apple tree images generated via prompt engineering, aiming to simulate real-world apple detection scenarios. The dataset creation process involved using the pre-trained Stable Diffusion model to generate images through carefully designed positive and negative prompts, and performing automatic annotation to identify the positions of apples. The primary application field of this dataset is apple detection in drone systems, to improve the efficiency of agricultural monitoring and infrastructure inspection.