Machine Learning Classification and Accuracy Assessment from High-Resolution Images of Coastal Wetlands

Prentice, Ricardo Martínez; Villoslada, Miguel; Ward, Raymond D.; Bergamo, Thaisa; Joyce, Christopher; Sepp, Kalev

doi:10.3390/rs13183669

Cited by 35 publications

(24 citation statements)

References 76 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, Samiappan et al [18] reported that when using the OBIA method based on UAV for coastal wetland mapping, extensive time and effort were spent in the selection of image segmentation parameters. In addition, Martínez Prentice et al [27] revealed that the PB method is more accurate than the OBIA method when monitoring coastal wetland vegetation based on UAV and suggested that the result of the PB method is more reflective of the distribution of vegetation in coastal wetlands with high landscape heterogeneity.…”

Section: Introductionmentioning

confidence: 99%

“…However, to our knowledge, almost no studies have evaluated the classification performance of the three paradigms, PB, OBIA, and DL, in coastal wetland UAV data. Since the PB and OBIA methods are the most common paradigm for UAV data processing in coastal wetlands [15,16,27] and are mutually independent of DL methods in processing and analysis, an understanding of the paradigm shift between them would be more practical and essential for coastal wetland monitoring.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Coastal Wetland Vegetation Classification Using Pixel-Based, Object-Based and Deep Learning Methods Based on RGB-UAV

Zheng

Hao

Wang

et al. 2022

Land

View full text Add to dashboard Cite

The advancement of deep learning (DL) technology and Unmanned Aerial Vehicles (UAV) remote sensing has made it feasible to monitor coastal wetlands efficiently and precisely. However, studies have rarely compared the performance of DL with traditional machine learning (Pixel-Based (PB) and Object-Based Image Analysis (OBIA) methods) in UAV-based coastal wetland monitoring. We constructed a dataset based on RGB-based UAV data and compared the performance of PB, OBIA, and DL methods in the classification of vegetation communities in coastal wetlands. In addition, to our knowledge, the OBIA method was used for the UAV data for the first time in this paper based on Google Earth Engine (GEE), and the ability of GEE to process UAV data was confirmed. The results showed that in comparison with the PB and OBIA methods, the DL method achieved the most promising classification results, which was capable of reflecting the realistic distribution of the vegetation. Furthermore, the paradigm shifts from PB and OBIA to the DL method in terms of feature engineering, training methods, and reference data explained the considerable results achieved by the DL method. The results suggested that a combination of UAV, DL, and cloud computing platforms can facilitate long-term, accurate monitoring of coastal wetland vegetation at the local scale.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Coastal Wetland Vegetation Classification Using Pixel-Based, Object-Based and Deep Learning Methods Based on RGB-UAV

Zheng

Hao

Wang

et al. 2022

Land

View full text Add to dashboard Cite

show abstract

“…Sentinel-2 data have been widely used to map different habitats and several studies have demonstrated their reliability in mapping and monitoring changes in marine biocenoses such as coral (Hedley et al 2012 ), mangrove (Pham et al 2019b , 2020 ) and seagrass beds Topouzelis et al 2016 ; Traganos and Reinartz 2018b ; Ha et al 2020 , 2021 ; Wicaksono et al 2021 ; Nur et al 2021 ; Ivajnšič et al 2022 ; Hartoni et al 2022 ). Many authors have used vegetation indices as the input for RF classification to map plant communities in terrestrial wetlands (Fletcher 2016 ) and coastal wetlands (Zoffoli et al 2020 ; Martínez Prentice et al 2021 ; Benmokhtar et al 2021 ), and compared to many machine learning algorithms and support vector machine techniques, the RF algorithm has produced promising results in terms of classifying seagrass (Zhang et al 2013 ; Traganos and Reinartz 2018b ; Ha et al 2020 ). For example, to monitor the dynamics of the Posidonia oceanica (L.) Delile meadows and Cymodocea nodosa (Ucria) Ascherson meadows in the Eastern Mediterranean, Traganos and Reinartz ( 2018a ) used the random forest algorithm and support vector machines for the RapidEye time series, after adjusting the atmospheric and analytical water column.…”

Section: Discussionmentioning

confidence: 99%

Monitoring the Spatial and Interannual Dynamic of Zostera noltei

et al. 2023

View full text Add to dashboard Cite

Seagrass is a vital structural and functional element of the marine environment worldwide and is highly valued for its ecological benefits. Monitoring the evolution of the seagrass habitat is essential to understand how this coastal ecosystem changes, and to develop good environmental management practices. For the present study, two remote sensing methods were used to map and monitor Zostera noltei Hornemann, 1832 ( Z. noltei ), in the Merja Zerga lagoon from 2010 to 2020. These methods which are the random forest algorithm and the object-oriented classification, were convenient to provide significant results. The first approach employed Sentinel-2 images from 2018 to 2020, which were used to extract information on changes in Z. noltei (commonly called dwarf eelgrass) distribution and aboveground biomass estimation. The second involved three orthophotography (orthophoto) mosaics from the years 2010, 2016, and 2018, which were analyzed to map the distribution of the species. It was revealed that Z. noltei coverage has increased by 212 ha since 2010, with most of the growth occurring in the center and upstream part of the lagoon. The mean aboveground biomass of dwarf eelgrass in the lagoon was 78.5 DW/m² in 2018, 92.6 DW/m² in 2019, and 115.2 g DW/m² in 2020. The approach used in this study has provided important insights into the dynamic and mean biomass of Z. noltei in the Merja Zerga lagoon. It is therefore a valuable, non-destructive method that uses freely-available Sentinel-2 satellite data.

show abstract

“…The spatial resolution of satellite imagery is sometimes inadequate to capture the spectral characteristics of the spatially variable vegetation types (Singhal et al, 2019). Thus, high-resolution unmanned aerial vehicles (UAVs) are gaining popularity among geospatial scientists, who recognize the benefits of their spectral information for spotting minute features (Martínez Prentice et al, 2021).…”

Section: Introductionmentioning

confidence: 99%

Vegetation discrimination based on chlorophyll prediction in Marshy wetland using Unmanned Aerial Vehicles

Mohanty,

Pandey,

Singh

et al. 2024

Aquatic Conservation

View full text Add to dashboard Cite

Wetlands are an integral part of our global ecosystems and play crucial roles in ecological functions such as carbon sequestration, flood mitigation, water purification, and recreational activities. The Ramsar Convention is the most significant wetland protection pact and is doing tremendous work in conserving wetlands worldwide. However, the wetlands area is still under threat due to anthropogenic activity. The current study utilized drone images, chlorophyll measurements and machine leaning to discriminate and map vegetation at marsh wetland area—the Ramsar site. The high‐resolution, multispectral imagery is acquired using a drone‐mounted MICAsense sensor. Eight spectral indices such as Normalized Difference Water Index (NDWI), Two‐Band Algorithm (2BDA), Normalized Difference Chlorophyll Index (NDCI), Normalized Difference Vegetation Index (NDVI) Enhanced Normalized Difference Vegetation Index (ENDVI), Green Normalized Difference Vegetation Index (GNDVI), and Normalised Difference RedEdge (NDRE) were calculated on the acquired imagery in order to discriminate the different vegetation covers such as floating aquatic vegetation (FAV), open water, and other vegetations types. These include the following: Eichhornia, Nymphea, Oleracea, Paspalam, and Oryza from agriculture land at the study site. Two models (viz., the Taylor plot and the Lek Profile methods) were employed to assess the sensitivity of the spectral indices for prediction of chlorophyll and vegetation discrimination. It is inferred from both methods that NDCI was most sensitive for chlorophyll prediction of vegetation followed by NGRDI/ ENDVI/ 2BDA and NDVI for chlorophyll prediction in wetland ecosystems. Further, three machine learning algorithms, support vector machine (SVM), random forest (RF), and gradient tree boost (GTB), were utilized for classification, and the performance accuracy of GTB was found to be the highest (0.893), followed by RF (0.851) and SVM (0.723). The GTB algorithm was applied over NDCI for vegetation discrimination. The study revealed that Eichhronia sp. is abundantly present at the study site; hence, strategic management plans should be carried out for the eradication of invasive species and proper management of wetland vegetation.

show abstract

Machine Learning Classification and Accuracy Assessment from High-Resolution Images of Coastal Wetlands

Cited by 35 publications

References 76 publications

Coastal Wetland Vegetation Classification Using Pixel-Based, Object-Based and Deep Learning Methods Based on RGB-UAV

Coastal Wetland Vegetation Classification Using Pixel-Based, Object-Based and Deep Learning Methods Based on RGB-UAV

Monitoring the Spatial and Interannual Dynamic of Zostera noltei

Vegetation discrimination based on chlorophyll prediction in Marshy wetland using Unmanned Aerial Vehicles

Contact Info

Product

Resources

About