How Useful is Region-based Classification of Remote Sensing Images in a Deep Learning Framework?

TL;DR

This study evaluates how different segmentation algorithms, especially superpixels, influence the accuracy of remote sensing image classification within a deep learning framework, highlighting the benefits of spatially-coherent segmentation.

Contribution

It demonstrates that superpixel segmentation improves classification accuracy by providing more homogeneous regions, enhancing deep learning-based remote sensing image classification.

Findings

Superpixel algorithms lead to higher classification accuracy.

Homogeneous, compact segments improve model generalization.

Segmentation quality significantly impacts deep learning classification performance.

Abstract

In this paper, we investigate the impact of segmentation algorithms as a preprocessing step for classification of remote sensing images in a deep learning framework. Especially, we address the issue of segmenting the image into regions to be classified using pre-trained deep neural networks as feature extractors for an SVM-based classifier. An efficient segmentation as a preprocessing step helps learning by adding a spatially-coherent structure to the data. Therefore, we compare algorithms producing superpixels with more traditional remote sensing segmentation algorithms and measure the variation in terms of classification accuracy. We establish that superpixel algorithms allow for a better classification accuracy as a homogenous and compact segmentation favors better generalization of the training samples.