Bayesian aggregation improves traditional single image crop classification approaches

TL;DR

This paper compares classical machine learning and neural network methods for crop classification from satellite images, demonstrating that Bayesian aggregation enhances accuracy over simple voting methods, with gradient boosting achieving 77.4% accuracy.

Contribution

It introduces Bayesian aggregation for field-wise crop classification from single satellite images, improving accuracy over traditional majority voting methods.

Findings

Bayesian aggregation improves classification accuracy by 1.5% over majority voting.

Gradient boosting achieves 77.4% overall accuracy and 0.66 macro F1-score.

Field-wise classification outperforms pixel-wise approaches.

Abstract

Machine learning (ML) methods and neural networks (NN) are widely implemented for crop types recognition and classification based on satellite images. However, most of these studies use several multi-temporal images which could be inapplicable for cloudy regions. We present a comparison between the classical ML approaches and U-Net NN for classifying crops with a single satellite image. The results show the advantages of using field-wise classification over pixel-wise approach. We first used a Bayesian aggregation for field-wise classification and improved on 1.5% results between majority voting aggregation. The best result for single satellite image crop classification is achieved for gradient boosting with an overall accuracy of 77.4% and macro F1-score 0.66.