Deep Neural Networks for Remote Sensing Applications

Dear Colleagues,

In recent years, several types of remotely sensed data, e.g., optical multi-spectral and hyper-spectral images, Synthetic Aperture Radar (SAR) and in many cases with extensive time series are increasingly available. In fact, several new Earth Online missions such as, the HyspIRI project developed by the NASA, the German EnMAP or the Italian PRISMA program, have increased the flow of generated remotely sensed data. All these remotely sensed data is motivating the development of large repositories and, most importantly, the development of advanced methods and algorithms for data analysis and processing.

On the other hand, Deep Neural Networks (DNNs), commonly called Deep Learning (DL) models, are showing very high potential in the recognition of spatial and temporal patterns in a wide range of remotely sensed applications (e.g., in scene classification, object detection, spectral unmixing, spatial super-resolution, pixel classification, dimensionality reduction, etc.), providing a great variety of algorithms, procedures and models, under different learning strategies (supervised, unsupervised, semi-supervised). In particular, Convolutional Neural Networks (CNNs), a type of DNNs, currently constitutes the-state-of-the art in image classification, object detection and instance segmentation. Generative adversarial networks (GANs) are showing promising results in the mapping of the terrestrial surface and in super-resolution problems. Recurrent Neural Networks (RNNs) are also showing good results in identifying patterns in time series and in forecasting meteorological events. 

However, due to the huge amount of parameters that need to be learned by DL models, the complex nature of DL models, the complexity of the remotely sensed data itself (e.g., high dimensionality) and the lack of labeled datasets, these approaches must deal with important problems, which can lead to inadequate generalization and loss of accuracy.

This Special Issue welcomes papers that explore novel and challenging applications by analyzing multi-band images acquired from diverse sensors using one or a combination of several Deep Learning models. We welcome topics that include but not limited to the following:

• Labeled dataset of unmanned aerial vehicle, aerial or satellite multi-band images;
• Dimensionality reduction;
• Feature extraction;
• Image fusion;
• Image reconstruction;
• Spectral unmixing methods;
• Remotely sensed image classification;
• Object detection in remotely sensed data, e.g., for designing new counting methods;
• Instance segmentation in remotely sensing data, e.g., for the delimitation of vegetation individuals, wildlife, catastrophic events, meteorological events;
• Adversarial generative networks, e.g., for super-resolution problems, classification of terrestrial surface, generation of synthesized images to fill gaps in clouds;
• Detection of anomalies in time series;
• Predictions, e.g., predicting specific climatic or polluting events, evolution of natural areas, changes in cities or crops;
• New problem applications addressed with deep learning models;

Dr. Siham Tabik

Dr. Emilio Guirado

Dr. Juan Mario Haut

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.