Soil Category Classification using Machine Vision and Automated NPK Content Estimation

Agriculture is the primary source of livelihood and income for the majority of people in India. Despite providing employment to 60% of India's population, agriculture only accounts for 20% of the country's GDP. This disparity might be the result of inadequate crop planning. To cope up with this problem, we propose an Android app-based solution using the Machine Vision approach. The app will take a picture of the soil and send it to the ML model for classification. NPK sensors are also used to measure soil nutrients, and the resulting values are uploaded to the cloud. By using NPK values and classified soil type, the system recommends the most suitable crop for particular soil type. The Support Vector Machine algorithm is utilized for both training and classification purposes and the Random Forest algorithm is used for crop prediction. After conducting extensive testing and validation, our proposed algorithm has demonstrated its ability to increase agricultural productivity. These findings underscore the potential of such technologies in revolutionizing modern agriculture by improving resource utilization, reducing environmental impact, and ultimately contributing to sustainable and resilient agroecosystems.