AgriVisio: An Intelligent Deep Learning Framework for Plant Disease Detection and Personalized Care Guidance

Abstract

Plant diseases significantly influence the crop production and agricultural sustainability. Early and appropriate user-friendly disease identification for personalized use for farmers is essential for effective crop management. Now-days advancement in deep learning and artificial intelligence afford promising results for automated plant disease detection from leaf images. In this study, we propose an integrated system- AgriVisio, that combines deep learning techniques for plant disease detection with real-time plant identification and customized care recommendations. We developed a custom Sequential Convolutional Neural Network (CNN) model trained on the Plant Village Dataset, available on Kaggle having fourteen types of plants (apple, blueberry, cheery, corn, grape, orange, peach, pepper, potato, raspberry, soyabean, squash, strawberry, tomato) suffering from either bacterial, viral or fungal infection of 38 type of diseases. Proposed AgriVisio model achieved an accuracy of 99.23%, significantly outperforming to identify plant disease and suggest care guidance, over pre-trained models. PlantID API is used to identify the species of the plant and OpenWeather API fetches real time weather data to provide location specific care guides and has potential improvement for agricultural practices from early disease detection and for directly providing personalized guidance.