Submit to Remote Sensing Review for Remote Sensing Propose a Special Issue

Journal Browser

AI Datasets, Tools, and Specifications for Earth Observation Applications

Special Issue Editors
Special Issue Information
Keywords
Published Papers

A special issue of Remote Sensing (ISSN 2072-4292). This special issue belongs to the section "Earth Observation Data".

Deadline for manuscript submissions: closed (15 January 2023) | Viewed by 13536

Share This Special Issue

Special Issue Editors

Dr. Liangcun Jiang

E-Mail Website
Guest Editor

School of Resources and Environmental Engineering, Wuhan University of Technology, Wuhan 430070, China
Interests: EO data infrastructure; data provenance; geospatial semantic web

Dr. Lei Hu

E-Mail Website
Guest Editor

Faculty of Resources and Environmental Science, Hubei University, 368 Youyi Road, Wuhan 430062, China
Interests: agro-geoinformatics; agricultural disasters; geospatial interoperability and standards; EO systems; GeoAI
Special Issues, Collections and Topics in MDPI journals

Prof. Dr. Peng Yue

E-Mail Website
Guest Editor

School of Remote Sensing and Information Engineering, Wuhan University, Wuhan 430079, China
Interests: Earth science data and information systems; GIS; data science; semantics; cloud computing
Special Issues, Collections and Topics in MDPI journals

Special Issue Information

Dear Colleagues,

In the past decade, Artificial Intelligence (AI) has become a research hotspot which brings great momentum to the area of Earth observation (EO). Various deep learning models have been and are being developed in the remote sensing community to harness the ever-increasing volume of EO data. However, compared with the widespread use of AI in computer vision systems (e.g., face recognition), the number of EO applications powered by AI is still very limited. One of the main bottlenecks is the lack of large-scale EO training datasets with high quality. New benchmark datasets with pixel-level, object-level, and scene-level labels, especially those involving multimodal RS images (e.g., multispectral, hyperspectral, and SAR data), are desperately needed to advance EO‐specific AI applications. Meanwhile, it is essential to make these training datasets findable, accessible, interoperable, and reusable (FAIR).

With this background in mind, in this Special Issue, we call for papers focusing on recent developments of datasets, specifications, and tools in EO applications using AI techniques. In particular, we encourage original research and review articles on methods for creating, collecting, describing, processing, analyzing, cataloging, sharing, and assessing EO training datasets. As EO data annotation is a laborious and time-consuming task, automatic annotation methods and open labeling tools are welcome as well. Potential topics include but are not limited to:

Multimodal RS benchmark datasets;
Open-source labeling tools/libraries for remote sensing images;
Automatic annotation of sample data;
Classification systems of RS label type;
Quality assessment of EO training datasets;
Specifications of sample data;
Specifications for GeoAI model interchange;
Data provenance;
Reproducibility in EO‐specific AI research;
EO data access and analysis through novel, standards-based techniques;
EO data applications and products using AI/ML techniques;
Knowledge-driven GeoAI.

Dr. Liangcun Jiang
Dr. Lei Hu
Prof. Dr. Peng Yue
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Remote Sensing is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2700 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

Keywords

EO training data
labeling tool
reproducibility
data specification

Published Papers (5 papers)

Download All Papers

Research

19 pages, 13419 KiB

Open AccessArticle

The Ground-Level Particulate Matter Concentration Estimation Based on the New Generation of FengYun Geostationary Meteorological Satellite

by Lin Tian, Lin Chen, Peng Zhang, Bo Hu, Yang Gao and Yidan Si

Remote Sens. 2023, 15(5), 1459; https://0-doi-org.brum.beds.ac.uk/10.3390/rs15051459 - 05 Mar 2023

Cited by 1 | Viewed by 1557

Abstract

The new-generation FengYun geostationary meteorological satellite has a high spatial and temporal resolution, which is advantageous in environmental assessments and air pollution monitoring. This study researched the ground-level particulate matter concentration estimation, based on satellite-observed radiations. The radiation of ground-level particulate matter is separate from the apparent radiation observed by satellites. The positive correlation between PM2.5 and PM10 is also considered to improve the accuracy of inversion results and the interpretability of the estimation model. Then, PM2.5 and PM10 concentrations were estimated synchronously every 5 min in mainland China based on FY-4A satellite directly observed radiations. The validation results showed that the improved model estimated results were close to the ground site measured results, with a high determination coefficient (R²) (0.89 for PM2.5, and 0.90 for PM10), and a small Root Mean Squared Error (RMSE) (4.69 μg/m³ for PM2.5 concentrations, and 13.77 μg/m³ for PM10 concentrations). The estimation model presented a good performance in PM2.5 and PM10 concentrations during typical haze and dust storm cases, indicating that it is applicable in different weather conditions and regions. Full article

(This article belongs to the Special Issue AI Datasets, Tools, and Specifications for Earth Observation Applications)

► Show Figures

Figure 1

32 pages, 7433 KiB

Open AccessArticle

FARMSAR: Fixing AgRicultural Mislabels Using Sentinel-1 Time Series and AutoencodeRs

by Thomas Di Martino, Régis Guinvarc’h, Laetitia Thirion-Lefevre and Elise Colin

Remote Sens. 2023, 15(1), 35; https://0-doi-org.brum.beds.ac.uk/10.3390/rs15010035 - 21 Dec 2022

Cited by 3 | Viewed by 1672

Abstract

This paper aims to quantify the errors in the provided agricultural crop types, estimate the possible error rate in the available dataset, and propose a correction strategy. This quantification could establish a confidence criterion useful for decisions taken on this data or to have a better apprehension of the possible consequences of using this data in learning downstream functions such as classification. We consider two agricultural label errors: crop type mislabels and mis-split crops. To process and correct these errors, we design a two-step methodology. Using class-specific convolutional autoencoders applied to synthetic aperture radar (SAR) time series of free-to-use and temporally dense Sentinel-1 data, we detect out-of-distribution temporal profiles of crop time series, which we categorize as one out of the three following possibilities: crop edge confusion, incorrectly split crop areas, and potentially mislabeled crop. We then relabel crops flagged as mislabeled using an Otsu threshold-derived confidence criteria. We numerically validate our methodology using a controlled disruption of labels over crops of confidence. We then compare our methods to supervised algorithms and show improved quality of relabels, with up to 98% correct relabels for our method, against up to 91% for Random Forest-based approaches. We show a drastic decrease in the performance of supervised algorithms under critical conditions (smaller and larger amounts of introduced label errors), with Random Forest falling to 56% of correct relabels against 95% for our approach. We also explicit the trade-off made in the design of our method between the number of relabels, and their quality. In addition, we apply this methodology to a set of agricultural labels containing probable mislabels. We also validate the quality of the corrections using optical imagery, which helps highlight incorrectly cut crops and potential mislabels. We then assess the applicability of the proposed method in various contexts and scales and present how it is suitable for verifying and correcting farmers’ crop declarations. Full article

(This article belongs to the Special Issue AI Datasets, Tools, and Specifications for Earth Observation Applications)

► Show Figures

Graphical abstract

19 pages, 65308 KiB

Open AccessArticle

Multi-Output Network Combining GNN and CNN for Remote Sensing Scene Classification

by Feifei Peng, Wei Lu, Wenxia Tan, Kunlun Qi, Xiaokang Zhang and Quansheng Zhu

Remote Sens. 2022, 14(6), 1478; https://0-doi-org.brum.beds.ac.uk/10.3390/rs14061478 - 18 Mar 2022

Cited by 17 | Viewed by 4628

Abstract

Scene classification is an active research area in the remote sensing (RS) domain. Some categories of RS scenes, such as medium residential and dense residential scenes, would contain the same type of geographical objects but have various spatial distributions among these objects. The adjacency and disjointness relationships among geographical objects are normally neglected by existing RS scene classification methods using convolutional neural networks (CNNs). In this study, a multi-output network (MopNet) combining a graph neural network (GNN) and a CNN is proposed for RS scene classification with a joint loss. In a candidate RS image for scene classification, superpixel regions are constructed through image segmentation and are represented as graph nodes, while graph edges between nodes are created according to the spatial adjacency among corresponding superpixel regions. A training strategy of a jointly learning CNN and GNN is adopted in the MopNet. Through the message propagation mechanism of MopNet, spatial and topological relationships imbedded in the edges of graphs are employed. The parameters of the CNN and GNN in MopNet are updated simultaneously with the guidance of a joint loss via the backpropagation mechanism. Experimental results on the OPTIMAL-31 and aerial image dataset (AID) datasets show that the proposed MopNet combining a graph convolutional network (GCN) or graph attention network (GAT) and ResNet50 achieves state-of-the-art accuracy. The overall accuracy obtained on OPTIMAL-31 is 96.06% and those on AID are 95.53% and 97.11% under training ratios of 20% and 50%, respectively. Spatial and topological relationships imbedded in RS images are helpful for improving the performance of scene classification. Full article

(This article belongs to the Special Issue AI Datasets, Tools, and Specifications for Earth Observation Applications)

► Show Figures

Figure 1

14 pages, 4197 KiB

Open AccessArticle

Combination of Models to Generate the First PAR Maps for Spain

by Francisco Ferrera-Cobos, Jose M. Vindel, Ousmane Wane, Ana A. Navarro, Luis F. Zarzalejo and Rita X. Valenzuela

Remote Sens. 2021, 13(23), 4950; https://0-doi-org.brum.beds.ac.uk/10.3390/rs13234950 - 06 Dec 2021

Cited by 1 | Viewed by 1844

Abstract

This work addresses the development of a PAR model in the entire territory of mainland Spain. Thus, a specific model is developed for each location of the study field. The new PAR model consists of a combination of the estimates of two previous models that had unequal performances in different climates. In fact, one of them showed better results with Mediterranean climate, whereas the other obtained better results under oceanic climate. Interestingly, the new PAR model showed similar performance when validated at seven stations in mainland Spain with Mediterranean or oceanic climate. Furthermore, all validation slopes ranged from 0.99 to 1.00; the intercepts were less than 3.70 μmol m⁻² s⁻¹; the R² were greater than 0.988, while MBE was closer to zero percent than −0.39%; and RMSE were less than 6.21%. The estimates of the PAR model introduced in this work were then used to develop PAR maps over mainland Spain that represent daily PAR averages of each month and a full year at all locations in the study field. Full article

(This article belongs to the Special Issue AI Datasets, Tools, and Specifications for Earth Observation Applications)

► Show Figures

Graphical abstract

24 pages, 7036 KiB

Open AccessArticle

Evaluation and Analysis of Poverty-Stricken Counties under the Framework of the UN Sustainable Development Goals: A Case Study of Hunan Province, China

by Yanjun Wang, Mengjie Wang, Bo Huang, Shaochun Li and Yunhao Lin

Remote Sens. 2021, 13(23), 4778; https://0-doi-org.brum.beds.ac.uk/10.3390/rs13234778 - 25 Nov 2021

Cited by 6 | Viewed by 2457

Abstract

Eliminating all forms of poverty in the world is the first United Nations Sustainable Development Goal (SDG). Developing a scientific and feasible method for monitoring and evaluating local poverty is important for the implementation of the SDG agenda. Based on the 2030 United Nations SDGs, in this paper, a quantitative evaluation model is built and applied to all poverty-stricken counties in Hunan Province. First, based on the SDG global index framework and local index system of China, a local SDG index system for poverty-related goals is designed, and the weights of the indexes are derived using an entropy method. The scores obtained for counties and districts with data available are then taken as the true value for the poverty assessment. Second, using National Polar-orbiting Visible Infrared Imaging Radiometer Suite (NPP-VIIRS) nighttime light images and land use and digital elevation model data, six factors, including socioeconomic, land cover, terrain and traffic factors, are extracted. Third, we then construct multiple linear evaluation models of poverty targets defined by the SDGs and machine learning evaluation models, including regression trees, support vector machines, Gaussian process regressions and ensemble trees. Last, combined with statistical data of poverty-stricken counties in Hunan Province, model validation and accuracy evaluation are carried out. The results show that the R² and relative error of the localized, multiple linear evaluation model, including all six factors, are 0.76 and 19.12%, respectively. The poverty-stricken counties in Hunan Province were spatially aggregated and distributed mainly in the southeastern and northwestern regions. The proposed method for regional poverty assessment based on multisource geographic data provides an effective poverty monitoring reference scheme for the implementation of the poverty eradication goals in the 2030 agenda. Full article

(This article belongs to the Special Issue AI Datasets, Tools, and Specifications for Earth Observation Applications)

► Show Figures

Figure 1

Journal Menu

Journal Browser

AI Datasets, Tools, and Specifications for Earth Observation Applications

Share This Special Issue

Special Issue Editors

Special Issue Information

Keywords

Published Papers (5 papers)

Research

Further Information

Guidelines

MDPI Initiatives

Follow MDPI