International Journal of Science, Engineering and TechnologyAuthors: Mrs. Divya C, Nandish K, Vishnu Prasad K, Dushyanth Gowda N, Nithin K.B
Abstract: In the current existing scenario, one of the main limitations to sustainable agricultural practices and crop production is the gradual degradation of soil quality. The expensive and time-consuming nature of traditional laboratory-based soil testing also poses a significant challenge for small- and marginal-scale farmers, consequently leading to inefficient soil utilization. In this study, Soil Intelligence 2.0- an artificial intelligence-based precision agricultural platform that leverages smartphone images for preliminary and real-time estimation of soil properties and cropping suggestion- is proposed to overcome this problem. The system integrates Vision Transformer (ViT) architecture with multimodal LLMs to evaluate the various soil visual properties such as texture, texture and pattern of the soil on an imageBased on these features, the system estimates important soil parameters such as nitrogen, phosphorus, potassium, and pH, and uses the estimated conditions to determine suitable crops. The platform further provides personalized agricultural recommendations through traditional, organic, and hybrid farming approaches. An AI-based agricultural assistant enables farmers to interact with the system and receive context-specific guidance related to crop selection, soil improvement, and cultivation practices. The proposed approach is designed to deliver preliminary soil insights within seconds using commonly available smartphones, offering a low-cost and accessible decision-support solution for precision farming. The system can assist farmers in making more informed field-level decisions while supporting efficient and sustainable agricultural practices.
DOI:
