The control of plant diseases is becoming more and more difficult; thus, effective and precise detection techniques are required. With an emphasis on the ResNet-18 architecture, this study offers a thorough analysis of the application of convolutional neural networks (CNNs) for the identification of plant diseases. Utilizing the PlantVillage dataset, which consists of more than 70,000 photos of 38 different illnesses in 16 different species, we improved the performance of the model by applying approaches for data augmentation and transfer learning. ResNet-18 outperformed other models such as VGG16, Inception V3, and Xception in terms of accuracy, achieving up to 99.8% in apple leaf disease detection, according to the comparative analysis. In order to facilitate user-submitted leaf picture diagnosis, the study also included establishing an end-to-end project environment and incorporating the learned model into a Django-based web application. The outcomes of the experiment showed high accuracy and usefulness, underscoring the possibility of implementation in the actual world. Future plans for extension include adding real-time monitoring for dynamic disease identification, investigating different architectures, and growing the dataset. The goal of this multidisciplinary strategy is to greatly increase agricultural yield by offering a dependable and easily available method for identifying plant diseases early on.