Scholarlink

Traditional crop mapping approaches, which rely primarily on manual field surveys, face limitations in scalability and efficiency, especially in intensive agricultural landscapes. Remote sensing (RS) provides an alternative with expansive spatial and temporal data acquisition capabilities. The complexity and volume of remote sensing data necessitate sophisticated analytical approaches. Therefore, this paper presents a novel sparse pixel-based training approach for Convolutional Neural Networks (CNNs) based on the ResNet architecture. The proposed approach comprises three principal phases: data preprocessing, classification, and evaluation. Seven distinct classes are employed to assess the effectiveness of the proposed method. Results indicate an impressive overall accuracy of 95%, precision of 94%, sensitivity of 95%, and f-score of 95%.

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

Deep Learning for Agricultural Crop and Land Cover Classification

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Mohamed A. Mostafa,
  • Anas M. Mansour,
  • Reem H. Aliraqi,
  • Ahmed Z. Elabdeen,
  • Moataz E. Sharaf,
  • Mahmoud S. Mohamed,
  • Menna T. Mahmoud,
  • Mariam M. Abdellatif,
  • Eman K. Elsayed

摘要

Traditional crop mapping approaches, which rely primarily on manual field surveys, face limitations in scalability and efficiency, especially in intensive agricultural landscapes. Remote sensing (RS) provides an alternative with expansive spatial and temporal data acquisition capabilities. The complexity and volume of remote sensing data necessitate sophisticated analytical approaches. Therefore, this paper presents a novel sparse pixel-based training approach for Convolutional Neural Networks (CNNs) based on the ResNet architecture. The proposed approach comprises three principal phases: data preprocessing, classification, and evaluation. Seven distinct classes are employed to assess the effectiveness of the proposed method. Results indicate an impressive overall accuracy of 95%, precision of 94%, sensitivity of 95%, and f-score of 95%.