Scholarlink

<p>Accurate and fast recognition of corn leaf disease has attracted increasing interest in recent years, which is of great significance for promoting corn production efficiency and quality. In this paper, we propose a progressive feature learning and multiplicative feature fusion network (PFMNet) for corn leaf disease recognition. Our approach leverages cloud services for efficient data storage, processing, and scalability. Specifically, we first develop a multi-scale feature learning module (MFLM) consisting of three parallel branches, which collaboratively learns the features from the input images, whole objects, and discriminative parts of objects in a global-to-local manner. This helps discover the subtle areas of lesions and suppress the complex background noise to enhance the feature discriminability, especially for the in-field scenes with overlapped, bent and distorted corn leaves. By mapping the features at different branches to a shared latent space and producing the channel attention for each branch, we further design a multiplicative feature fusion module (MFFM), which helps aggregate different scale features in a synergistic way. Extensive experimental results quantitatively and qualitatively demonstrate the effectiveness of the proposed method, which can achieve an accuracy of 97.14% and recall of 97.36% on the corn leaf disease recognition benchmark. The integration of cloud services enhances the scalability and practicality of our solution for widespread deployment in agricultural applications. The code is available at <a href="https://github.com/JunlingWang0227/PFMNet.">https://github.com/JunlingWang0227/PFMNet.</a></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Progressive feature learning and multiplicative feature fusion network for corn disease recognition

  • 期刊论文

  • 发表日期：

    2026年7月1日

查看全文
  • Junling Wang,
  • Hua Fang,
  • Wei Wei,
  • Ping Zong,
  • Can Xu

摘要

Accurate and fast recognition of corn leaf disease has attracted increasing interest in recent years, which is of great significance for promoting corn production efficiency and quality. In this paper, we propose a progressive feature learning and multiplicative feature fusion network (PFMNet) for corn leaf disease recognition. Our approach leverages cloud services for efficient data storage, processing, and scalability. Specifically, we first develop a multi-scale feature learning module (MFLM) consisting of three parallel branches, which collaboratively learns the features from the input images, whole objects, and discriminative parts of objects in a global-to-local manner. This helps discover the subtle areas of lesions and suppress the complex background noise to enhance the feature discriminability, especially for the in-field scenes with overlapped, bent and distorted corn leaves. By mapping the features at different branches to a shared latent space and producing the channel attention for each branch, we further design a multiplicative feature fusion module (MFFM), which helps aggregate different scale features in a synergistic way. Extensive experimental results quantitatively and qualitatively demonstrate the effectiveness of the proposed method, which can achieve an accuracy of 97.14% and recall of 97.36% on the corn leaf disease recognition benchmark. The integration of cloud services enhances the scalability and practicality of our solution for widespread deployment in agricultural applications. The code is available at https://github.com/JunlingWang0227/PFMNet.