• Version
• Download 19
• File Size 423.29 KB
• File Count 1
• Create Date 23 July 2025
• Last Updated 23 July 2025

CNN-Based Tomato Leaf Detection and ANN-Driven Fertilizer Recommendations

1ERUSU. KATA RAJU REDDY, 2SHAIK SURI BABU

1Assistant Professor, 22MCA Final Semester,

1Master of Computer Applications,

1Sanketika Vidya Parishad Engineering College, Vishakhapatnam, Andhra Pradesh, India

Abstract

This thesis introduces an AI-based system for improving tomato crop management by combining two neural networks: a Convolutional Neural Network (CNN) for leaf disease detection and a Feed-forward Neural Network (FNN) for fertilizer recommendation. The CNN processes 128×128×3 RGB images of tomato leaves and classifies them into ten categories (nine diseases and one healthy class) using techniques like separable convolutions and attention mechanisms for high accuracy and low complexity. Data augmentation and normalization boost its generalization, achieving over 90% accuracy. The FNN recommends fertilizers based on the identified disease, soil parameters (pH, nitrogen, phosphorus, potassium), and environmental factors (humidity, temperature), reaching near-100% accuracy even with limited data. Built using TensorFlow and Keras, the system is optimized for real-time deployment on edge devices like Raspberry Pi, ensuring accessibility in rural areas. Testing confirmed high accuracy, quick inference times, and robustness. Although minor misclassifications exist in similar disease types, and the FNN can benefit from more diverse data, future improvements such as transfer learning and mobile integration are planned to enhance usability and impact in precision agriculture

IndexTerms: Tomato Leaf Disease Detection, Convolutional Neural Network (CNN), Fertilizer Recommendation System, Feedforward Neural Network (FNN), Precision Agriculture, Edge Computing (Raspberry Pi).