Scholarlink

Agricultural farming is a cornerstone of society. Population growth demands a sustainable food source with ever-increasing yield demands. However, there are many challenges with large-scale industrial farming. The relatively low biodiversity of crops, alongside crops often being selected for yield volume rather than their robustness, leaves them susceptible to rapid plant disease spread. Timely detection and treatment can help improve crop yields and prevent crop failures. However, continuous monitoring at a suitable detail is impractical. The use of artificial intelligence, specifically in the field of computer vision has yet to be explored for crop monitoring. High-resolution cameras and suitably optimized convolutional models can help detect developing conditions. This work proposed a computer vision-based approach, optimized by a modified metaheuristic in order to attain the best possible performance. The proposed approach is meticulously tested on publicly available data with the best models exceeding an accuracy of 83.47% suggesting the practical viability of the introduced methodology.

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

Optimizing Convolutional Networks for Plant Disease Detection Applications Using a Modified Optimizer

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Stefan Ivanovic,
  • Luka Jovanovic,
  • Angelina Njegus,
  • Zivkovic Miodrag,
  • Branislav Radomirovic,
  • Nebojsa Bacanin

摘要

Agricultural farming is a cornerstone of society. Population growth demands a sustainable food source with ever-increasing yield demands. However, there are many challenges with large-scale industrial farming. The relatively low biodiversity of crops, alongside crops often being selected for yield volume rather than their robustness, leaves them susceptible to rapid plant disease spread. Timely detection and treatment can help improve crop yields and prevent crop failures. However, continuous monitoring at a suitable detail is impractical. The use of artificial intelligence, specifically in the field of computer vision has yet to be explored for crop monitoring. High-resolution cameras and suitably optimized convolutional models can help detect developing conditions. This work proposed a computer vision-based approach, optimized by a modified metaheuristic in order to attain the best possible performance. The proposed approach is meticulously tested on publicly available data with the best models exceeding an accuracy of 83.47% suggesting the practical viability of the introduced methodology.