International Journal of Science, Engineering and TechnologyAuthors: Kishan kannaujiya, Sandeep Bind, Kundan Kumar, Naimisha Awasthi
Abstract: Accurate and early detection of Leaf diseases is a critical component of precision agriculture, as it helps in reducing crop losses and ensuring sustainable food production. Traditional deep learning–based disease detection systems rely exclusively on leaf image analysis and neglect environmental and soil factors that influence disease occurrence and treatment outcomes. This study presents an advanced, real-time, integrated decision- support system that combines Convolutional Neural Networks (CNN) for leaf disease identification, soil health analysis using pH and NPK parameters, and real-time weather data integration through the OpenWeather API to enhance treatment precision and crop management. The proposed system operates in three major phases: (i) Leaf image preprocessing and classification using a fine- tuned CNN model trained on the PlantVillage dataset, achieving 98.6A rule-based treatment recommendation engine integrates the CNN classification output with soil and weather context to generate actionable and environment-specific suggestions, including suitable fertilizers, pesticides, and preventive measures. The system also incorporates a disease severity estimation module that quantifies infection extent and a risk prediction model that estimates short-term outbreak probabilities using recent climatic data. Experimental evaluation demonstrates that integrating soil and weather data improves treatment recommendation accuracy by 18Overall, the integration of deep learning with contextual environmental intelligence provides a comprehensive, data-driven, and farmer-friendly solution that supports decision-making, reduces input waste, and promotes sustainable agricultural practices. This framework can be further extended for large-scale deployment in smart farming environments through IoT and drone-based automation.
DOI: http://doi.org/10.5281/zenodo.17750325
