Deep Learning for Remote Sensing Image Scene Classification

Dear Colleagues,

Remote sensing image scene classification plays a crucial role in understanding and interpreting satellite, aerial, and terrestrial imagery. Traditionally, machine learning techniques like Support Vector Machine (SVM) and Random Forest (RF) were used for this task, but these methods often relied on hard-crafted features and could struggle with the vast variability in remote sensing data. The advent of deep learning has revolutionized this field by offering automated feature extraction and enhanced classification performance, especially in handling complex and large-scale datasets.

Deep learning models, particularly Convolutional Neural Networks (CNNs), have become the dominant approach for remote sensing image scene classification. CNNs are well suited for image analysis due to their ability to learn spatial hierarchies of features automatically. By using multiple layers, CNNs capture both low-level features like edges and textures and higher-level semantic features, making them highly effective in distinguishing between different land use and land cover types.

One of the primary advantages of deep learning in this domain is its ability to generalize across diverse scenes. This is critical in remote sensing where images vary widely in resolution, perspective, and spectral properties.

The aim of this Special Issue on Deep Learning for Remote Sensing Image Scene Classification is to bring together cutting-edge research and advancements in the application of deep learning techniques to remote sensing imagery. With the growing volume and complexity of remote sensing data, deep learning has emerged as a transformative approach, offering unprecedented capabilities in automated feature extraction, scalability, and classification accuracy.

This Special Issue seeks to:

1. Highlight recent breakthroughs in deep learning: showcase the state-of-the-art deep learning algorithms and models, such as CNNs, Recurrent Neural Networks (RNNs), and transformer-based architectures, and how they are being adapted to enhance scene classification in remote sensing.
2. Encourage interdisciplinary collaboration: Promote the exchange of ideas between experts in deep learning, remote sensing, computer vision, and environmental sciences. The goal is to foster collaborations that push the boundaries of traditional image classification techniques and create innovative solutions to complex challenges.
3. Address key challenges in remote sensing data: tackle issues such as data heterogeneity, high dimensionality, and the scarcity of labeled datasets in remote sensing, exploring how deep learning models can be trained, fine-tuned, and optimized to handle these challenges effectively.
4. Promote innovation in data processing and model design: Highlight recent innovations in data augmentation, transfer learning, and big data processing that are critical for handling the vast and diverse nature of remote sensing imagery. Emphasis will be placed on methods that improve model robustness, efficiency, and generalization across different sets of data.
5. Expand applications of scene classification: Explore new and emerging applications of deep learning for remote sensing, such as land use and land cover classification for digital twins, disaster monitoring and prevention, sustainable urban planning, and environmental management. The focus will be on how deep learning enables more accurate, scalable, and efficient analysis in these areas.

The topics of interest for this Special Issue are as follows:

• Advanced deep learning architectures;
• Exploration of hybrid models that combine multiple deep learning techniques to enhance classification performance;
• Transfer learning and domain adaptation;
• Role of synthetic data and simulation in training deep learning models when real-world labeled datasets are limited;
• Multispectral and hyperspectral image classification;
• Challenges of spectral variability and high dimensionality in remote sensing data;
• Spatio-temporal analysis;
• Application of deep learning for monitoring environmental changes, urban dynamics, and disaster evolution over time;
• Big data and cloud computing;
• Scalable deep learning solutions;
• Parallelization and optimization of learning workflows;
• Explainability and interpretability using deep learning to offer insights into the decision-making process of scene classification
• Techniques to improve transparency and reliability of deep learning models in critical applications such as disaster management or environmental monitoring;
• Emerging applications of deep learning in remote sensing, such as precision agriculture, smart cities, and autonomous systems;
• Exploration of how deep learning-based scene classification can be integrated with other geospatial technologies.

Prof. Dr. Christine Pohl
Guest Editor

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.

• deep learning
• remote sensing
• image scene classification
• convolutional neural networks (CNNs)
• transfer learning
• multispectral and hyperspectral imaging
• data augmentation
• big data analytics
• smart city
• digital twins