Identification of growth stages of crops using mobile phone images and machine learning

Abstract

Determination of crop’s growth stages greatly helps farmers in understanding adaptability of several crops in arable land. The growth stages of crops play an important role in forecasting crop’s yield. Monitoring crop growth is a challenging task due to the fact that it is lengthy, difficult and time-consuming activity. Therefore, an easy and reliable mechanism of identification of various growth stages of crop is highly needed. The purpose of this study is to determine growth stages of chili and cotton crop via regular mobile phone images. Pakistan is one of the world’s top cotton producers, while Sindh province is the world’s leading chili producer. Machine learning approach is used in developing a prediction model for growth stages identification. The model is trained using six supervised machine learning algorithms, namely Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), K-Nearest Neighbor (K-NN), Gaussian Naive Bayes (NB) and logistic regression (LR). The experiments provided twofold information: 1) identification of best machine learning algorithm for this task and 2) significantly accurate forecasting of chili and cotton crop growth stages. The findings of the study may help in smart agriculture for better productivity and timely getting in-depth information of crops.