您好，欢迎访问安徽省农业科学院 机构知识库！

Causality-inspired crop pest recognition based on Decoupled Feature Learning

文献类型： 外文期刊

作者： Hu, Tao 1 ; Du, Jianming 2 ; Yan, Keyu 1 ; Dong, Wei 3 ; Zhang, Jie 2 ; Wang, Jun 4 ; Xie, Chengjun 2 ;

作者机构： 1.Univ Sci & Technol China, Sci Isl Branch, Grad Sch, Hefei, Peoples R China

2.Chinese Acad Sci, Inst Intelligent Machines, Hefei Inst Phys Sci, Hefei 230031, Peoples R China

3.Anhui Acad Agr Sci, Agr Econ & Informat Res Inst, Hefei 230001, Peoples R China

4.Anhui Tech Coll Mech & Elect Engn, Sch Internet & Telecommun, Wuhu, Peoples R China

关键词： pest recognition; Decoupled Feature Learning; causal inference; deep learning

期刊名称：PEST MANAGEMENT SCIENCE （ 影响因子：3.8； 五年影响因子：4.3 ）

ISSN： 1526-498X

年卷期： 2024 年

页码：

收录情况： SCI

摘要： BACKGROUNDEnsuring the efficient recognition and management of crop pests is crucial for maintaining the balance in global agricultural ecosystems and ecological harmony. Deep learning-based methods have shown promise in crop pest recognition. However, prevailing methods often fail to address a critical issue: biased pest training dataset distribution stemming from the tendency to collect images primarily in certain environmental contexts, such as paddy fields. This oversight hampers recognition accuracy when encountering pest images dissimilar to training samples, highlighting the need for a novel approach to overcome this limitation.RESULTSWe introduce the Decoupled Feature Learning (DFL) framework, leveraging causal inference techniques to handle training dataset bias. DFL manipulates the training data based on classification confidence to construct different training domains and employs center triplet loss for learning class-core features. The proposed DFL framework significantly boosts existing baseline models, attaining unprecedented recognition accuracies of 95.33%, 92.59%, and 74.86% on the Li, DFSPD, and IP102 datasets, respectively.CONCLUSIONExtensive testing on three pest datasets using standard baseline models demonstrates the superiority of DFL in pest recognition. The visualization results show that DFL encourages the baseline models to capture the class-core features. The proposed DFL marks a pivotal step in mitigating the issue of data distribution bias, enhancing the reliability of deep learning in agriculture. (c) 2024 Society of Chemical Industry. We propose the Decoupled Feature Learning framework to enhance the performance of deep learning models in pest recognition. This framework mitigates training dataset bias by employing causal inference to construct distinct training domains and utilizes center triplet loss to encourage the model to learn class-core features. image

  • 相关文献
作者其他论文 更多>>
置顶
购物车
反馈

© 京ICP备09089781号-29 主办单位：安徽省农业科学院

学术资源： 中国农业科学院 农业专业知识服务系统 农科联盟资源共建共享平台 中国农业科技文献与信息服务平台 中国农业期刊集成服务平台