AI-Powered Crop Disease Detection: An Integrated Approach to Plant Disease and Insect Classification

Effective management of plant diseases and pests is crucial for ensuring agricultural productivity and minimizing economic losses. Traditional detection methods are often slow, labor-intensive, and prone to inaccuracies. This paper presents an AI-based system that leverages deep learning to accurately detect plant diseases and classify pests from images. The proposed approach utilizes the VGG16 architecture with transfer learning, trained on two distinct datasets from Kaggle: one containing approximately 87,000 RGB images across 38 plant disease classes and another comprising about 3100 RGB images for 9 different pests. Image preprocessing techniques, including resizing to 224×224 pixels and normalization, were applied to improve model performance. A newer approach for training the disease detection models is used. VGG16 model was adapted by replacing the fully connected layers with custom dense layers tailored for classification tasks. The system was trained using the Adam optimizer and evaluated on metrics such as accuracy, precision, recall, and F1-score, demonstrating high performance in both disease and pest classification. A user-friendly web and mobile interface enable users to upload or capture images for real-time analysis. Upon detection, the system provides detailed reports, treatment suggestions, and direct purchase links for recommended pesticides. Future developments aim to incorporate drone integration, and advanced models like Vision Transformers (ViTs) to enhance accuracy, precision and scalability. This research highlights the potential of AI-driven approaches in advancing precision agriculture and promoting sustainable farming practices.