A Comparative Analysis of Deep Learning-Based Segmentation Techniques for Terrain Classification in Aerial Imagery

Background: Deep convolutional neural networks (CNNs) have become widely popular for many imaging applications, and they have also been applied in various studies for monitoring and mapping areas of land. Nevertheless, most of these networks were designed to perform in different scenarios, such as a...

Full description

Saved in:
Bibliographic Details
Main Authors: Martina Formichini, Carlo Alberto Avizzano
Format: Article
Language:English
Published: MDPI AG 2025-07-01
Series:AI
Subjects:
Online Access:https://www.mdpi.com/2673-2688/6/7/145
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850067499989598208
author Martina Formichini
Carlo Alberto Avizzano
author_facet Martina Formichini
Carlo Alberto Avizzano
author_sort Martina Formichini
collection DOAJ
description Background: Deep convolutional neural networks (CNNs) have become widely popular for many imaging applications, and they have also been applied in various studies for monitoring and mapping areas of land. Nevertheless, most of these networks were designed to perform in different scenarios, such as autonomous driving and medical imaging. Methods: In this work, we focused on the usage of existing semantic networks applied to terrain segmentation. Even though several existing networks have been used to study land segmentation using transfer learning methodologies, a comparative analysis of how the underlying network architectures perform has not yet been conducted. Since this scenario is different from the one in which these networks were developed, featuring irregular shapes and an absence of models, not all of them can be correctly transferred to this domain. Results: Fifteen state-of-the-art neural networks were compared, and we found that, in addition to slight differences in performance, there were relevant differences in the numbers and types of outliers that were worth highlighting. Our results show that the best-performing models achieved a pixel-level class accuracy of 99.06%, with an F1-score of 72.94%, 71.5% Jaccard loss, and 88.43% recall. When investigating the outliers, we found that PSPNet, FCN, and ICNet were the most effective models. Conclusions: While most of this work was performed on an existing terrain dataset collected using aerial imagery, this approach remains valid for investigation of other datasets with more classes or richer geographical extensions. For example, a dataset composed of Copernicus images opens up new opportunities for large-scale terrain analysis.
format Article
id doaj-art-f3443bf13e8e48ebb82bf96db9f8d5b9
institution DOAJ
issn 2673-2688
language English
publishDate 2025-07-01
publisher MDPI AG
record_format Article
series AI
spelling doaj-art-f3443bf13e8e48ebb82bf96db9f8d5b92025-08-20T02:48:17ZengMDPI AGAI2673-26882025-07-016714510.3390/ai6070145A Comparative Analysis of Deep Learning-Based Segmentation Techniques for Terrain Classification in Aerial ImageryMartina Formichini0Carlo Alberto Avizzano1Institute of Mechanical Intelligence, Via Alamanni, 13b, 56010 Ghezzano, Pisa, ItalyInstitute of Mechanical Intelligence, Via Alamanni, 13b, 56010 Ghezzano, Pisa, ItalyBackground: Deep convolutional neural networks (CNNs) have become widely popular for many imaging applications, and they have also been applied in various studies for monitoring and mapping areas of land. Nevertheless, most of these networks were designed to perform in different scenarios, such as autonomous driving and medical imaging. Methods: In this work, we focused on the usage of existing semantic networks applied to terrain segmentation. Even though several existing networks have been used to study land segmentation using transfer learning methodologies, a comparative analysis of how the underlying network architectures perform has not yet been conducted. Since this scenario is different from the one in which these networks were developed, featuring irregular shapes and an absence of models, not all of them can be correctly transferred to this domain. Results: Fifteen state-of-the-art neural networks were compared, and we found that, in addition to slight differences in performance, there were relevant differences in the numbers and types of outliers that were worth highlighting. Our results show that the best-performing models achieved a pixel-level class accuracy of 99.06%, with an F1-score of 72.94%, 71.5% Jaccard loss, and 88.43% recall. When investigating the outliers, we found that PSPNet, FCN, and ICNet were the most effective models. Conclusions: While most of this work was performed on an existing terrain dataset collected using aerial imagery, this approach remains valid for investigation of other datasets with more classes or richer geographical extensions. For example, a dataset composed of Copernicus images opens up new opportunities for large-scale terrain analysis.https://www.mdpi.com/2673-2688/6/7/145convolutional neural networksdeep learningremote sensingcomputer visionprecision agriculture
spellingShingle Martina Formichini
Carlo Alberto Avizzano
A Comparative Analysis of Deep Learning-Based Segmentation Techniques for Terrain Classification in Aerial Imagery
AI
convolutional neural networks
deep learning
remote sensing
computer vision
precision agriculture
title A Comparative Analysis of Deep Learning-Based Segmentation Techniques for Terrain Classification in Aerial Imagery
title_full A Comparative Analysis of Deep Learning-Based Segmentation Techniques for Terrain Classification in Aerial Imagery
title_fullStr A Comparative Analysis of Deep Learning-Based Segmentation Techniques for Terrain Classification in Aerial Imagery
title_full_unstemmed A Comparative Analysis of Deep Learning-Based Segmentation Techniques for Terrain Classification in Aerial Imagery
title_short A Comparative Analysis of Deep Learning-Based Segmentation Techniques for Terrain Classification in Aerial Imagery
title_sort comparative analysis of deep learning based segmentation techniques for terrain classification in aerial imagery
topic convolutional neural networks
deep learning
remote sensing
computer vision
precision agriculture
url https://www.mdpi.com/2673-2688/6/7/145
work_keys_str_mv AT martinaformichini acomparativeanalysisofdeeplearningbasedsegmentationtechniquesforterrainclassificationinaerialimagery
AT carloalbertoavizzano acomparativeanalysisofdeeplearningbasedsegmentationtechniquesforterrainclassificationinaerialimagery
AT martinaformichini comparativeanalysisofdeeplearningbasedsegmentationtechniquesforterrainclassificationinaerialimagery
AT carloalbertoavizzano comparativeanalysisofdeeplearningbasedsegmentationtechniquesforterrainclassificationinaerialimagery