Scholarlink

UAV Image plays an vital role in the earlier prediction of plant diseases as it contains more spectral information. However, in the UAV images, multi-leaf disease prediction in the multi-plants is unfocused research. Hence, by using UAV data, this article proposes a multi-plant multi-disease prediction framework. Primarily, the target regions (i.e. leaf) are identified based on the Jeffries-Matusita-based Simple Linear Iterative Clustering (JM-SLIC) segmentation in the input UAV images. Next, with dead pixel replacement and noise removal, the segmented image is pre-processed. Then, the Stochastic gradient-based Bi-cubic Interpolation (S-BCI) technique is used to approximate the resolution. Next, resolution-approximated spectral images are unmixed, and the chlorophyll content-based indexes are estimated. By using Radial Convolution-based FractalNet (RC-FNet), the indexes are combined with the features to predict the diseases in the leaf. In the end, the proposed framework’s disease prediction efficiency is proved by the experimental evaluation.

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

Multi-plant Disease Prediction Using Radial Convolution-Based FractalNet with Resolution Approximated UAV Images

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • D. Lita Pansy,
  • M. Murali

摘要

UAV Image plays an vital role in the earlier prediction of plant diseases as it contains more spectral information. However, in the UAV images, multi-leaf disease prediction in the multi-plants is unfocused research. Hence, by using UAV data, this article proposes a multi-plant multi-disease prediction framework. Primarily, the target regions (i.e. leaf) are identified based on the Jeffries-Matusita-based Simple Linear Iterative Clustering (JM-SLIC) segmentation in the input UAV images. Next, with dead pixel replacement and noise removal, the segmented image is pre-processed. Then, the Stochastic gradient-based Bi-cubic Interpolation (S-BCI) technique is used to approximate the resolution. Next, resolution-approximated spectral images are unmixed, and the chlorophyll content-based indexes are estimated. By using Radial Convolution-based FractalNet (RC-FNet), the indexes are combined with the features to predict the diseases in the leaf. In the end, the proposed framework’s disease prediction efficiency is proved by the experimental evaluation.