Paper Title
Plant Diseases Detection using Deep Learning

Abstract
Agricultural productivity is often affected by the plant diseases that reduce both yield and quality of crops. Conventional disease identification practices depend on largely on visual inspection by specialists, a process that becomes inefficient due to the time required and the need for domain-specific expertise. To overcome this limitation, the present study investigates a deep learning–based framework for the automated detection and classification of diseases affecting plant leaves. A ResNet-34 model initialized with pre- trained weights was selected in this study to classify multiple crop diseases using the PlantVillage dataset comprising more than 50,000 images. To enhance the model’s ability to generalize, data augmentation strategies were introduced to simulate variations in lighting and background conditions. An overall testing accuracy of 96.21% was recorded, reflecting reliable performance across multiple disease categories. Grad- CAM was further applied to visualize the regions influencing the model’s predictions, allowing clearer interpretation of infected leaf areas. The proposed method shows potential for integration into mobile and web-based platforms, enabling real-time disease diagnosis and promoting sustainable agricultural practices. Keywords - Deep Learning Applications, Plant Disease Classification, CNN Architectures, ResNet-34, Transfer Learning, Grad-CAM Visualization, Precision Agriculture.