Detection and classification of nutrient deficiencies in plants using machine learning

Anu Jose,S Nandagopalan,Dhanya Viswanath,Vidya Ubalanka

doi:10.1088/1742-6596/1850/1/012050

Anu Jose, S Nandagopalan + Show 2 more

Open Access

https://doi.org/10.1088/1742-6596/1850/1/012050

Copy DOI

Abstract

Agriculture is the major factor contributing to Indian Economy. According to the current statistics, its contribution to GDP sector is 17.9%. Technical advancement in agricultural domain will produce more agricultural products without any wastage of money, time and manpower. Nutrients play a major role in plant growth. Lack of nutrients leads to reduced crop yield and plant growth. In this work, we are trying to create an artificial neural network model to recognize and classify the nutrient deficiency in tomato by examining the leaf characteristics. This will help farmers to adjust the nutrient supply to the plant. If soil lacks a specific nutrient, it will reflect in the physical characteristics of a leaf. The color and shape of a leaf are the two major features used for identifying the nutrient deficiency. The comparison of different segmentation schemes like hue based and threshold based schemes shows their influence in the performance of the proposed system. The influence of different activation functions in the artificial neural network is also studied in this work. The results show that the proposed method was able to classify and identify nutritional deficiencies with high accuracy.

Highlights

According to the statistics of 2019, India is the second largest tomato producer in the world after china
Figure16: Receiver Operating Characteristic (ROC) Curve for the classifier Different activation functions like RELU, Softmax and SWISH are applied in neural network .The accuracy comparison is given in the below diagram.SWISH will worl better than RELU and softmax for the multi label classification
Scope of Work An image processing model consisting of features from hue is the best model for the task of tomato leaf classification

Summary

Introduction

According to the statistics of 2019, India is the second largest tomato producer in the world after china. The major symptoms of nutrient deficiencies are leaf yellowing; brown spots an on leaf and stem This will leads to stunted growth and poor flowering and fruiting. Nitrogen Nitrogen is the major requirement for protein creation for tomato plant It will increase the yield and helps to achieve proper growth. Several images of tomato leaves with deficiencies like nitrogen, potassium, phosphorous, calcium, Sulfur and magnesium are used to train the artificial neural network classifier. The major objectives of the work includes1) Collect image data set of various common tomato diseases caused by macro nutrients deficiency. Boron deficiency in corn plant is recognized by Luz et al, [5] by inspecting the RGB images of leaves It was applying KNN algorithm for classification. Another scheme is for histogram based segmentation, which is usually applicable for gray scale images

Hue Based segmentation

Findings

Conclusion and Future