A Cloud-Based Framework for Large-Scale Monitoring of Ocean Plastics Using Multi-Spectral Satellite Imagery and Generative Adversarial Network

Jamali, Ali; Mahdianpari, Masoud

doi:10.3390/w13182553

Cited by 19 publications

(20 citation statements)

References 67 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As discussed, generating new data in remote sensing, specifically wetland mapping, is time-consuming, labor-intensive, and costly. On the other hand, although the DL methods have been successfully employed in different fields of remote sensing, such as object detection [45], [46], [47], [48], [49] and classification [50], [51], [52], [53], [54], [55], they have a need for a large set of training data [3]. This issue can be addressed with the utilization of the new architecture of GANs that revolutionized the DL field introduced by Goodfellow et al [56].…”

Section: Generative Adversarial Networkmentioning

confidence: 99%

3-D Hybrid CNN Combined With 3-D Generative Adversarial Network for Wetland Classification With Limited Training Data

Jamali¹,

Mahdianpari

Mohammadimanesh³

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

Self Cite

View full text Add to dashboard Cite

Recently, deep learning algorithms, specifically convolutional neural networks (CNNs), have played an important role in remote sensing image classification, including wetland mapping. However, one limitation of deep CNN for classification is its requirement for a great number of training samples. This limitation is particularly enhanced when the classes of interest are spectrally similar, such as that of wetland types, and the training samples are limited. This article presents a novel approach named 3-D hybrid generative adversarial network (3-D hybrid GAN) that addresses the limited training sample issue in the classification of remote sensing imagery with a focus on complex wetland classification. We used a conditional map unit that generates synthetic training samples for only classes with a lower number of training samples to improve the per-class accuracy of wetlands. This procedure overcomes the issue of imbalanced data in conventional wetland mapping. Based on the achieved results, better classification accuracy is obtained by integrating a 3-D generative adversarial network (3-D GAN) and the CNN network of a 3-D hybrid CNN using both 3-D and 2-D convolutional filters. Experimental results on the avalon pilot site located in eastern Newfoundland, Canada, and covering five wetland types of bog, fen, marsh, swamp, and shallow water demonstrate that our model significantly outperforms other CNN models, including the HybridSN, SpectralNet, MLP-mixer, as well as a conventional algorithm of random forest for complex wetland classification by approximately 1% to 51% in terms of F-1 score.

show abstract

Section: Generative Adversarial Networkmentioning

confidence: 99%

3-D Hybrid CNN Combined With 3-D Generative Adversarial Network for Wetland Classification With Limited Training Data

Jamali¹,

Mahdianpari

Mohammadimanesh³

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

Self Cite

View full text Add to dashboard Cite

show abstract

“…For improving wetland mapping in Canada, with the utilization of different data sources and techniques, there has been extensive research by different research groups [45,46]. For instance, Jamali et al [47] used Sentinel-1 and Sentinel-2 data for the classification of five wetlands of bog, fen, marsh, swamp, and shallow water in Newfoundland, Canada, with the use of very deep CNN networks and the Generative Adversarial Network (GAN) [48][49][50][51], reaching a high average accuracy of 92.30%. In their research, creating synthetic samples of Sentinel-1 and Sentinel-2 data significantly improved the classification accuracy of wetland mapping.…”

Section: Related Workmentioning

confidence: 99%

“…The advantage of a deep learning classifier is that over time and with their advancement, they become more capable of complex scene classification with less training data. For example, synthetic wetland training data can be produced by GAN networks [47,48] to overcome the biggest disadvantage of deep learning methods. Moreover, multi-model deep learning classifiers can be developed to obtain higher classification accuracies [53,54].…”

Section: Related Workmentioning

confidence: 99%

Swin Transformer for Complex Coastal Wetland Classification Using the Integration of Sentinel-1 and Sentinel-2 Imagery

Jamali

Mahdianpari

2022

Water

Self Cite

View full text Add to dashboard Cite

The emergence of deep learning techniques has revolutionized the use of machine learning algorithms to classify complicated environments, notably in remote sensing. Convolutional Neural Networks (CNNs) have shown considerable promise in classifying challenging high-dimensional remote sensing data, particularly in the classification of wetlands. State-of-the-art Natural Language Processing (NLP) algorithms, on the other hand, are transformers. Despite the fact that transformers have been utilized for a few remote sensing applications, they have not been compared to other well-known CNN networks in complex wetland classification. As such, for the classification of complex coastal wetlands in the study area of Saint John city, located in New Brunswick, Canada, we modified and employed the Swin Transformer algorithm. Moreover, the developed transformer classifier results were compared with two well-known deep CNNs of AlexNet and VGG-16. In terms of average accuracy, the proposed Swin Transformer algorithm outperformed the AlexNet and VGG-16 techniques by 14.3% and 44.28%, respectively. The proposed Swin Transformer classifier obtained F-1 scores of 0.65, 0.71, 0.73, 0.78, 0.82, 0.84, and 0.84 for the recognition of coastal marsh, shrub, bog, fen, aquatic bed, forested wetland, and freshwater marsh, respectively. The results achieved in this study suggest the high capability of transformers over very deep CNN networks for the classification of complex landscapes in remote sensing.

show abstract

“…As a result, the scope and focus of this study are to review the literature on using satellites to monitor macro-, meso-, micro-, and nanoplastics in water. Furthermore, hyperspectral, thermal imaging, and multispectral sensors have been successfully applied to monitoring macroplastic debris [15,16,29,[35][36][37][38][39][40][41][42] , and researchers recently discovered that SAR can be used for monitoring microplastics, a research domain still in its infancy [24] . The significance of monitoring global plastic regardless of size, occurrence, source, and location is emphasized by many researchers [42,50] .…”

Section: The Scope Of the Studymentioning

confidence: 99%

“…As a result, it is challenging or impossible for human brains and traditional statistical methods to perform pattern analysis on satellite data in order to establish relationships between variables of interest. Accordingly, machine and deep learning (ML/DL) techniques have found useful applications in monitoring plastic pollution in aquatic systems [35][36][37][38][39][40][41][42][43] . Artificial intelligence (AI)-processed satellites are transforming the manner in which experts study plastic pollution in global aquatic ecosystems, and they are proving to be a new powerful tool in the fight to protect water systems from pollution.…”

Section: Introductionmentioning

confidence: 99%

Satellite sensors as an emerging technique for monitoring macro- and microplastics in aquatic ecosystems

Mukonza¹,

Chiang²

2022

Water Emerg Contam & Nanoplastics

View full text Add to dashboard Cite

Plastic pollution in aquatic ecosystems has been identified as a growing global water pollution threat that is negatively impacting water quality and, as a result, affecting the health of humans, aquatic animals, and wildlife. Therefore, it presents a global environmental catastrophe that requires immediate attention. Plastics in water (in their different forms, macro-, meso-, micro-, and nanoplastics) are contaminants of emerging concerns that have since evolved to be a global environmental threat. Despite increasing levels of pollution in aquatic ecosystems, there are insufficient monitoring data to evaluate the extent of the catastrophe. Traditional methods of monitoring plastics in water are constrained by high sampling costs, intensive labor, and limited temporal and spatial coverage, which results in limited monitoring data. Thus, insufficient monitoring data limit our understanding of the true quantities and persistence of plastic particles in aquatic ecosystems, as well as the extent to which they impact the aquatic environment. There is increasing availability of free big geospatial data (amounting to petabytes/day) from satellite sensors for potentially monitoring plastics. This provides a possible solution to these challenges by minimizing the fieldwork required and therefore reducing the costs and sampling time. The study purpose of this review is to analyze advances in emerging technology such as the use of satellite sensors to monitor the occurrence of macro- and microplastics in freshwater, ultimately aimed at creating new operational monitoring solutions. This review: (1) examines the literature to identify trends, accomplishments, and limitations of using satellite data to monitor plastics in water; (2) identifies and compares traditional, and machine and deep learning satellite image classification methods for monitoring plastics in water; and (3) identifies research gaps and summarizes future perspectives and recommendations to improve monitoring methods.

show abstract

A Cloud-Based Framework for Large-Scale Monitoring of Ocean Plastics Using Multi-Spectral Satellite Imagery and Generative Adversarial Network

Cited by 19 publications

References 67 publications

3-D Hybrid CNN Combined With 3-D Generative Adversarial Network for Wetland Classification With Limited Training Data

3-D Hybrid CNN Combined With 3-D Generative Adversarial Network for Wetland Classification With Limited Training Data

Swin Transformer for Complex Coastal Wetland Classification Using the Integration of Sentinel-1 and Sentinel-2 Imagery

Satellite sensors as an emerging technique for monitoring macro- and microplastics in aquatic ecosystems

Contact Info

Product

Resources

About