Optimizing Satellite Imagery Datasets for Enhanced Land/Water Segmentation

This paper presents an automated procedure for optimizing datasets used in land/water segmentation tasks with deep learning models. The proposed method employs the Normalized Difference Water Index (NDWI) with a variable threshold to automatically assess the quality of annotations associated with mu...

Full description

Saved in:
Bibliographic Details
Main Authors: Marco Scarpetta, Luisa De Palma, Attilio Di Nisio, Maurizio Spadavecchia, Paolo Affuso, Nicola Giaquinto
Format: Article
Language:English
Published: MDPI AG 2025-03-01
Series:Sensors
Subjects:
Online Access:https://www.mdpi.com/1424-8220/25/6/1793
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850087916122931200
author Marco Scarpetta
Luisa De Palma
Attilio Di Nisio
Maurizio Spadavecchia
Paolo Affuso
Nicola Giaquinto
author_facet Marco Scarpetta
Luisa De Palma
Attilio Di Nisio
Maurizio Spadavecchia
Paolo Affuso
Nicola Giaquinto
author_sort Marco Scarpetta
collection DOAJ
description This paper presents an automated procedure for optimizing datasets used in land/water segmentation tasks with deep learning models. The proposed method employs the Normalized Difference Water Index (NDWI) with a variable threshold to automatically assess the quality of annotations associated with multispectral satellite images. By systematically identifying and excluding low-quality samples, the method enhances dataset quality and improves model performance. Experimental results on two different publicly available datasets—the SWED and SNOWED—demonstrate that deep learning models trained on optimized datasets outperform those trained on baseline datasets, achieving significant improvements in segmentation accuracy, with up to a 10% increase in mean intersection over union, despite a reduced dataset size. Therefore, the presented methodology is a promising scalable solution for improving the quality of datasets for environmental monitoring and other remote sensing applications.
format Article
id doaj-art-96cc1cfdfea745fdb03cedb4c9c9e76d
institution DOAJ
issn 1424-8220
language English
publishDate 2025-03-01
publisher MDPI AG
record_format Article
series Sensors
spelling doaj-art-96cc1cfdfea745fdb03cedb4c9c9e76d2025-08-20T02:43:07ZengMDPI AGSensors1424-82202025-03-01256179310.3390/s25061793Optimizing Satellite Imagery Datasets for Enhanced Land/Water SegmentationMarco Scarpetta0Luisa De Palma1Attilio Di Nisio2Maurizio Spadavecchia3Paolo Affuso4Nicola Giaquinto5Department of Electrical and Information Engineering, Polytechnic University of Bari, Via E. Orabona 4, 70125 Bari, ItalyDepartment of Electrical and Information Engineering, Polytechnic University of Bari, Via E. Orabona 4, 70125 Bari, ItalyDepartment of Electrical and Information Engineering, Polytechnic University of Bari, Via E. Orabona 4, 70125 Bari, ItalyDepartment of Electrical and Information Engineering, Polytechnic University of Bari, Via E. Orabona 4, 70125 Bari, ItalyDepartment of Electrical and Information Engineering, Polytechnic University of Bari, Via E. Orabona 4, 70125 Bari, ItalyDepartment of Electrical and Information Engineering, Polytechnic University of Bari, Via E. Orabona 4, 70125 Bari, ItalyThis paper presents an automated procedure for optimizing datasets used in land/water segmentation tasks with deep learning models. The proposed method employs the Normalized Difference Water Index (NDWI) with a variable threshold to automatically assess the quality of annotations associated with multispectral satellite images. By systematically identifying and excluding low-quality samples, the method enhances dataset quality and improves model performance. Experimental results on two different publicly available datasets—the SWED and SNOWED—demonstrate that deep learning models trained on optimized datasets outperform those trained on baseline datasets, achieving significant improvements in segmentation accuracy, with up to a 10% increase in mean intersection over union, despite a reduced dataset size. Therefore, the presented methodology is a promising scalable solution for improving the quality of datasets for environmental monitoring and other remote sensing applications.https://www.mdpi.com/1424-8220/25/6/1793dataset quality evaluationmetrology for AINDWIwater detectioncoastline monitoringremote sensing
spellingShingle Marco Scarpetta
Luisa De Palma
Attilio Di Nisio
Maurizio Spadavecchia
Paolo Affuso
Nicola Giaquinto
Optimizing Satellite Imagery Datasets for Enhanced Land/Water Segmentation
Sensors
dataset quality evaluation
metrology for AI
NDWI
water detection
coastline monitoring
remote sensing
title Optimizing Satellite Imagery Datasets for Enhanced Land/Water Segmentation
title_full Optimizing Satellite Imagery Datasets for Enhanced Land/Water Segmentation
title_fullStr Optimizing Satellite Imagery Datasets for Enhanced Land/Water Segmentation
title_full_unstemmed Optimizing Satellite Imagery Datasets for Enhanced Land/Water Segmentation
title_short Optimizing Satellite Imagery Datasets for Enhanced Land/Water Segmentation
title_sort optimizing satellite imagery datasets for enhanced land water segmentation
topic dataset quality evaluation
metrology for AI
NDWI
water detection
coastline monitoring
remote sensing
url https://www.mdpi.com/1424-8220/25/6/1793
work_keys_str_mv AT marcoscarpetta optimizingsatelliteimagerydatasetsforenhancedlandwatersegmentation
AT luisadepalma optimizingsatelliteimagerydatasetsforenhancedlandwatersegmentation
AT attiliodinisio optimizingsatelliteimagerydatasetsforenhancedlandwatersegmentation
AT mauriziospadavecchia optimizingsatelliteimagerydatasetsforenhancedlandwatersegmentation
AT paoloaffuso optimizingsatelliteimagerydatasetsforenhancedlandwatersegmentation
AT nicolagiaquinto optimizingsatelliteimagerydatasetsforenhancedlandwatersegmentation