Land Use Classification in Remote Sensing Images by Convolutional Neural Networks

TL;DR

This paper demonstrates that convolutional neural networks, especially fine-tuned pre-trained models, significantly improve the accuracy of land use classification in remote sensing images across diverse datasets.

Contribution

It introduces the use of pre-trained CNN architectures like CaffeNet and GoogLeNet for remote sensing scene classification, showing their effectiveness and broad applicability.

Findings

Pre-trained CNNs outperform training from scratch.

Fine-tuning reduces overfitting and design time.

Significant performance gains over state-of-the-art methods.

Abstract

We explore the use of convolutional neural networks for the semantic classification of remote sensing scenes. Two recently proposed architectures, CaffeNet and GoogLeNet, are adopted, with three different learning modalities. Besides conventional training from scratch, we resort to pre-trained networks that are only fine-tuned on the target data, so as to avoid overfitting problems and reduce design time. Experiments on two remote sensing datasets, with markedly different characteristics, testify on the effectiveness and wide applicability of the proposed solution, which guarantees a significant performance improvement over all state-of-the-art references.