Machine Learning in Indian Crop Classification of Temporal Multi-Spectral Satellite Image

Recently, there has been a remarkable growth in Artificial Intelligence (AI) with the development of efficient AI models and high-power computational resources for processing complex datasets. There have been a growing number of applications of machine learning in satellite remote sensing image data processing. In India, agriculture has a huge impact on the national economy and most of the critical decisions are dependent on agricultural statistics. In this work, machine learning models have been applied for crop classification of Sentinel-2 satellite temporal remote sensing image data. Guntur district region of Andhra Pradesh, India has been used as the study area. The main reasons for selecting this region are the diversity of agricultural crops and the availability of ground truth. The performance of machine learning models Support Vector Machine (SVM), Random Forest (RF), Convolution Neural Network (CNN), Recurrent Neural Network (RNN) with Long Short-Term Memory (LSTM) and RNN with Gated Recurrent Unit (GRU) have been evaluated for crop classification. Classification accuracies are generally evaluated by using test data. In most cases the classification accuracy from test data is not commensurate to estimated crop areas from the classified image. Such methods limit the estimated crop areas acceptance for official purposes. The uniqueness of this work is the classification accuracy is evaluated by estimated crop areas. The results show that SVM has the best F1 score of 0.99 and estimated major crop areas have 95.9% agreement with the ground surveyed crop area.