Scholarlink

Pigeon pea is a significant pulse crop in India, valued for its high protein content, but its production is hindered by leaf diseases that cause substantial economic losses. To address these challenges, this paper presented an automatic categorization for identifying pigeon pea leaf disorders using deep learning techniques. The proposed structure involves of three phases: preprocessing of image data, leaf area segmentation, and sickness classification. A lightweight U-Net model is introduced to accurately extract the leaf region from the background. Experimental results demonstrate the effectiveness of the segmentation, with an Intersection over Union (IoU) score of 0.92 and a similarity index of 0.94, outperforming state-of-the-art segmentation methods. A customized CNN model is employed for classification, and it has a 95% overall accuracy rate in identifying leaf illnesses. The proposed approach shows potential for improving agriculture output and disease control for the production of pigeon pea in India and other countries.

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

Pigeon Pea Leaf Region Extraction for Classification Using Modified U-Net and CNN Architectures

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • G. G. Rajput,
  • Vanita Bhimappa Doddamani

摘要

Pigeon pea is a significant pulse crop in India, valued for its high protein content, but its production is hindered by leaf diseases that cause substantial economic losses. To address these challenges, this paper presented an automatic categorization for identifying pigeon pea leaf disorders using deep learning techniques. The proposed structure involves of three phases: preprocessing of image data, leaf area segmentation, and sickness classification. A lightweight U-Net model is introduced to accurately extract the leaf region from the background. Experimental results demonstrate the effectiveness of the segmentation, with an Intersection over Union (IoU) score of 0.92 and a similarity index of 0.94, outperforming state-of-the-art segmentation methods. A customized CNN model is employed for classification, and it has a 95% overall accuracy rate in identifying leaf illnesses. The proposed approach shows potential for improving agriculture output and disease control for the production of pigeon pea in India and other countries.