PlantNet: Scalable Convolutional Neural Network for Image-Based Plant Disease Detection

Plant diseases significantly impact global agricultural productivity, necessitating the development of reliable, efficient, and scalable diagnostic systems for timely intervention and yield protection. This research presents PlantNet, a novel Convolutional Neural Network (CNN) architecture tailored for accurate identification of plant diseases from images. By leveraging deep learning techniques, PlantNet processes large-scale image datasets to detect disease symptoms with high precision. The model employs transfer learning, utilizing pre-trained networks on vast image repositories before fine-tuning on a specialized plant disease dataset, thereby enhancing feature extraction while minimizing computational complexity. The architecture is composed of multiple convolutional and pooling layers, ensuring both depth and performance efficiency. To improve the model's generalizability in real-world conditions, data augmentation techniques—such as random rotations, shifts, and lighting adjustments—are applied to address variations in orientation, illumination, and background noise. The dataset used encompasses a diverse collection of plant species and associated diseases, offering a robust foundation for training and validation. Experimental results demonstrate that PlantNet outperforms existing approaches in early disease detection, achieving an average precision of 95% and recall of 93%>. This performance enables prompt identification and management of plant health issues, contributing to improved crop resilience and food security. Furthermore, the system's scalability makes it suitable for deployment across various agricultural environments, including mobile-based and edge computing applications. Overall, PlantNet demonstrates the potential of deep learning-based image analysis in advancing automated plant disease diagnostics, offering a promising tool for precision agriculture and sustainable farming practices in both developed and developing regions.