ABSTRACT
This paper presents an algorithm for detecting micronutrients present in the soil using Image Processing and Convolutional Neural Network technique. Soil test is necessary because excessive or lacking use of manure will destroy the crop plant. Huge quantity and less quantity of fertilizer can affect the crop yield. Earlier, farmer used traditional methods to grow the crops to prepare their own farm in the traditional method which they had learnt from their intimates without noticing the micronutrients present in the soil that can change from time to time. This lead to low productivity because the farmer's did not have the proper understanding of the soil. Nowadays, there are several methods to analyze the soil sample through soil test laboratories but these methods are time consuming and tidous. However, Image Processing and Convolutional Neural Network is used efficiently to determine the micronutrients and pH level of soil i.e., Zinc(Zn), Iron(Fe), pH, Manganese(Mn) and Copper(Cu). The soil images were captured manually from Udupi Soil Health Lab. The system is divided into five sections i.e., Image capturing, Image Pre-Processing, Image Segmentation, Training System and Result. Soil images are captured using Smartphone and are provided as an input to the system, from the input images HSV values (Hue Saturation Value) are extracted and these HSV values(Hue Saturation Value) are given as an input to Convolutional Neural Network and as a result soil micronutrients value of the input image is displayed. Based on the result, the soil micronutrients and Accuracy of the model is calculated and displayed. The dataset is split into training and testing wherein, 80% of data is used to train the model and 20% of data was used for testing the model. This project is implemented on MATLAB. The maximum accuracy of the model obtained was 95%.
