Plant Leaf plays an important role in the first identification and accurate classification of diseases to ensure crop health, reduce agricultural loss and support food security in agricultural communities. we present a deep learning -based system for detecting and classifying plant diseases using a specially designed deep convolutional neural network. Unlike the pre-trained transmission models that require high calculation resources, which were developed from scratch to provide real-time, effective and accurate predictions suitable for mobile and online purposes. The model was trained and validated using a publicly available plant inguinal data sets, including thousands of healthy and diseased plant leaves from many crop types, including tomatoes, potatoes, corn and more. Our approach includes image presentation technology, computer text and waiver-based regularization to increase model generalization on unseen data. The trained model gained more than 98% accuracy on the verification set and demonstrated strong classification performance in different types of plants. To improve access and purpose, an online gradio interface was integrated, allowing users to upload leaf images and the disease name, throne levels, symptoms, precautions and proposed treatment, as well as predictions. It covered the gap between the user-focused interface machine learning research and practical agricultural application. The proposed solution is adapted to placement in the resource environment and has a significant ability to help farmers, agricultural workers and decision makers in the monitoring of the disease and crop management. Integration of AI-driven agriculture not only reduces the dependence on manual inspection but also contributes to a large measure of permanent agriculture and aids to improve multilingual support from partially infected or noisy leaf images.