Machine Learning and Deep Learning for Soil Analysis and Classification of Micro and Macro Nutrient Using IOT

Abstract The soil is the entity that keeps living on Earth alive. Despite substantial progress in the service sector, agriculture remains India's primary source of employment and revenue. The soil sample is a useful method for determining the present nutrient level of soil and determining the appropriate quantity of nutrition to apply to a soil depending on its fertility requirements.Finding the availability of seeds, evaluating the need for crops in the market, watching the soil, weather, and water resources, and choosing an acceptable crop based on these aspects are all crucial in agriculture.There have been a lot of developments lately, ranging from crop selection to crop cutting. The Internet of Things, cloud computing, and machine learning techniques primarily assist farmers in analysing and improving their decision-making at every step of production. He should also have the ability to decide wisely at every level of farming. The decision support system must use artificial intelligence, machine learning, the cloud, sensors, and other automated devices in order to deliver the correct information quickly. To suggest crops, we have put forth an Internet of Things-enabled approach called IoTSNA-CR (soil nutrient classification and crop recommendation model). In order to improve production, the model assists in minimising the use of fertilisers to the soil.The suggested methodology is divided into stages, such as gathering real-time data from agricultural areas using IoT sensors and storing it in cloud.Then after that, pre-processing data and doing recurring analysis on it with various learning strategies.Different sensors, including a pH, GPS, water level indication, soil temperature, soil moisture, and colour sensor, were included in a cost-effective sensory system that was assembled.We were able to gather data on moisture, temperature, water level, soil NPK colour values, date, time, longitude, and latitude thanks to this sensing system.The purpose of,this effort is,to,look at the major soil characteristics that influence crop growth, such,as organic,matter, important plant,nutrients, major nutrients, and,micronutrients, and use Machine,Learning,and Deep,Learning,models to classify soil fertility. To determine which region of soil is better, ML and DL models are employed in intra-class soil classification. Major and micronutrients are included in the dataset. Iron (Fe), Manganese(Mn), Zinc(Zn), Boron(B), and Copper(Cu) are micronutrient elements, whereas Organic carbon(OC), Nitrogen(N), Phosphorus Pentoxide(P2O5), and Potassium oxide (K2O) are major nutrition elements. Soil testing is an important technique for determining the.available.nutrient.status.ofsoil.and.the.appropriate.quantity.of.nutrients.to.be.applied.to.a.specific.soil.depending on its fertility and crop demands. The soil experiment report results are used to categorize numerous important soil properties such as soil,fertility.indices of.present Organic,Carbon(OC), Iron(Fe), and Manganese(Mn). The long,short-term,memory,network (LSTM) and Artificial Neural Network were used to create a deep learning model. For soil classification, ML models,such,as a KNN, SVM,and,RF techniques used. The performance of the Deep Learning model, which achieves about 98 percent accuracy, outperforms that of the Machine Learning model. Some issues need to be resolved to further enhance the performance of deep learning models in solving problems related to soil classification. The dataset has a big influence on performance. To improve the training process and the performance of deep learning models, consider focusing on the production of a well-established dataset that is relevant to the real-world scenario.