The widespread impact of plant diseases on agricultural yield demands an intelligent, accessible, and real-time detection solution. This paper presents the Mobile Application for Automated Plant Disease Detection, a lightweight smartphone-based system that leverages a fine-tuned MobileNetV2 Convolutional Neural Network (CNN) to classify plant leaf diseases in real time. The system accepts smartphone camera images, applies preprocessing including resizing to 224×224 pixels and normalisation, and classifies the plant health state as Healthy, Early Blight, Late Blight, Leaf Curl, or Powdery Mildew with associated confidence scores. A Flask API backend hosts the trained model and communicates with a React Native mobile frontend to return classification results within 1.2 seconds on a 4G network connection. Firebase Cloud Messaging delivers real-time push notifications and treatment recommendations directly to the farmer's device. The system is deployed entirely on standard Android and iOS smartphones without any specialised hardware, sensors, or wearable devices. Experimental evaluation on the PlantVillage dataset with over 54,000 annotated leaf images demonstrated classification accuracy exceeding 96%, API response latency below 800 milliseconds, and zero dependency on dedicated agricultural equipment. Usability testing with agricultural practitioners confirmed intuitive operation without prior technical training. These results confirm that the proposed application offers an efficient, portable, and institutionally deployable solution for modern precision agriculture.

Keywords—Plant disease detection; MobileNetV2; convolutional neural network; deep learning; precision agriculture; smartphone application; transfer learning; PlantVillage dataset; real-time classification; push notification