Agro-analysis: Soil Classification and Crop Matching with ML Models

Abstract

Dying soil, poor crop yields, and wasteful use of resources are long-term issues in agriculture, particularly in underdeveloped regions. This research provides a framework that integrates soil picture categorization with crop compatibility guidelines powered by machine learning. The method is primarily based on Artificial Intelligence (AI). Black, yellow, peat, laterite, and cinder soils are categorized via the device’s use of SoilNet, a bespoke Convolutional Neural Network (CNN). After the type manner is whole, a hybrid recommendation engine is used to forecast the quality plants to develop using parameters including soil pH, temperature, humidity, and rainfall. This engine incorporates models from deep learning, decision tree, random forest, and Support Vector Machine (SVM). To enhance user involvement and operational transparency, the solution is carried out as a web-based utility, which is strengthened with a Telegram bot. This bot provides real-time schooling updates. The system’s robustness and realistic software are proven experimentally with a soil class accuracy of 92.10% on a curated dataset. Through using synthetic intelligence, this method promotes sustainable agricultural practices and precision agriculture by imparting practical insights into soil fitness and crop planning. Farmers and other agricultural stakeholders can benefit significantly from the counseled system’s ability to improve statistics-driven decision-making.