Scholarlink

Accurate and timely classification of tomato leaf diseases is crucial for effective disease management and food security. Traditional methods relying on expert knowledge and manual inspection are time-consuming, subjective, and often inaccurate. Advanced models achieve high accuracy but are limited by high computational demands and poor generalizability across crops and diseases. Furthermore, dependence on sophisticated hardware can limit practical application in less accessible regions. This paper bridges these gaps by developing a highly efficient and precise hybrid model for the classification of several diseases in tomato plants. The proposed model integrates DenseNet201 and InceptionV3, utilizing feature fusion and fine tuning to enhance performance. The experimental setup involves a comprehensive dataset of over 32,000 classified images of tomato leaves, augmented to increase size and variability. The hybrid model excelled through feature fusion and fine-tuning, achieving a high accuracy of 98.5%, precision of 97.3%, recall of 97.8%, and an F1 score of 97.6%. The study highlights the hybrid model’s potential for practical agricultural use, offering an adaptable solution for farmers and researchers.

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

Enhancing Tomato Leaf Disease Classification Integrating DenseNet201 and InceptionV3 Models with Fine Tuning and Feature Fusion

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Abhiram Sharma,
  • R. Srivats,
  • S. Abirami,
  • T. Nathezhtha

摘要

Accurate and timely classification of tomato leaf diseases is crucial for effective disease management and food security. Traditional methods relying on expert knowledge and manual inspection are time-consuming, subjective, and often inaccurate. Advanced models achieve high accuracy but are limited by high computational demands and poor generalizability across crops and diseases. Furthermore, dependence on sophisticated hardware can limit practical application in less accessible regions. This paper bridges these gaps by developing a highly efficient and precise hybrid model for the classification of several diseases in tomato plants. The proposed model integrates DenseNet201 and InceptionV3, utilizing feature fusion and fine tuning to enhance performance. The experimental setup involves a comprehensive dataset of over 32,000 classified images of tomato leaves, augmented to increase size and variability. The hybrid model excelled through feature fusion and fine-tuning, achieving a high accuracy of 98.5%, precision of 97.3%, recall of 97.8%, and an F1 score of 97.6%. The study highlights the hybrid model’s potential for practical agricultural use, offering an adaptable solution for farmers and researchers.