摘要
苹果叶部状态是反映苹果长势良好与否的晴雨表,及时准确地识别苹果叶病类型并采取防治措施对于保证苹果的产量和质量十分重要。不同时间采集的叶部数据样本往往因存在环境的变化,造成采样数据分布的差异,使得传统深度学习方法不能很好地解决跨域识别问题。本研究提出一种基于领域自适应的苹果叶病识别方法,通过对齐不同领域的数据分布,实现了苹果叶病的跨域分类识别。实验结果验证了本研究方法的有效性,为农业领域的跨域识别问题提供了新的研究思路。
The state of apple leaves is a barometer of whether the apple is growing well or not.It is very important to identify the types of apple leaf disease timely and accurately and take control measures to ensure the yield and quality of apple.The leaf data samples collected at different times often have environmental changes,resulting in differences in the distribution of sampling data,so that the traditional deep learning methods can not solve the problem of cross-domain recognition.In this study,a domain-adaptive apple leaf disease recognition method is proposed,which realizes the cross-domain classification and recognition of apple leaf disease by aligning the data distribution in different fields.The experimental results verify the effectiveness of this research method,and provide a new research idea for cross-domain identification in the field of agriculture.
出处
《智慧农业导刊》
2021年第12期9-11,共3页
JOURNAL OF SMART AGRICULTURE
基金
泰山学院引进人才科研启动基金项目资助(编号:Y-01-2018006)。
关键词
深度学习
领域自适应
苹果叶病识别
deep learning
domain adaptation
apple leaf disease recognition
