Scholarlink

This paper delves into a novel scenario where disease symptoms remain latent in plants, rendering traditional visual detection methods inadequate. Despite the potential advantages of using machine learning or deep learning algorithms for plant disease diagnosis based on symptoms, inherent limitations arise. One such limitation lies in the reliance on visible symptoms. Particularly, in cases where plants are in the early stages of disease development, or during what can be termed as a “window period,” physical symptoms may not manifest visibly on the plant's foliage. This poses a significant challenge in accurately identifying and classifying diseases using symptom-based approaches. The absence of visible cues makes it difficult for conventional machine learning models to discern the presence of diseases accurately. Consequently, relying solely on visible symptoms may lead to false negatives, where diseased plants are incorrectly classified as healthy due to the lack of observable manifestations. In order to address the limitations posed by the reliance on visible symptoms for disease detection in plants, spectral-based methodologies were used to construct the dataset. Spectral analysis involves the measurement of electromagnetic radiation emitted or reflected by objects, which can provide valuable insights into their biochemical composition and physiological status.

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

Spectral-Based Mango Disease Classification: Introduction of a Novel Dataset

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Aishwariya Budhrani,
  • Hardikkumar Jayswal,
  • Axat Patel,
  • Naina Parmar,
  • Maulik Shah,
  • Rajesh Patel,
  • Jaimin Undaviya,
  • Ashwin Makvana

摘要

This paper delves into a novel scenario where disease symptoms remain latent in plants, rendering traditional visual detection methods inadequate. Despite the potential advantages of using machine learning or deep learning algorithms for plant disease diagnosis based on symptoms, inherent limitations arise. One such limitation lies in the reliance on visible symptoms. Particularly, in cases where plants are in the early stages of disease development, or during what can be termed as a “window period,” physical symptoms may not manifest visibly on the plant's foliage. This poses a significant challenge in accurately identifying and classifying diseases using symptom-based approaches. The absence of visible cues makes it difficult for conventional machine learning models to discern the presence of diseases accurately. Consequently, relying solely on visible symptoms may lead to false negatives, where diseased plants are incorrectly classified as healthy due to the lack of observable manifestations. In order to address the limitations posed by the reliance on visible symptoms for disease detection in plants, spectral-based methodologies were used to construct the dataset. Spectral analysis involves the measurement of electromagnetic radiation emitted or reflected by objects, which can provide valuable insights into their biochemical composition and physiological status.