Mohamed Fawzy
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Mohamed Fawzy has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Árpád Barsi
Department of Photogrammetry and Geoinformatics, Faculty of Civil Engineering, Budapest University of Technology and Economics, Műegyetem rkp. 3, H-1111 Budapest, Hungary
4 shared publications - Attila Juhasz
Department of Photogrammetry and Geoinformatics, Faculty of Civil Engineering, Budapest University of Technology and Economics, Műegyetem rkp. 3, H-1111 Budapest, Hungary
1 shared publication - Karem Abdelmohsen
School of Sustainability Arizona State University Tempe AZ USA
1 shared publication - Lóránt Földváry
Department of Geodesy and Surveying, Faculty of Civil Engineering, Budapest University of Technology and Economics, Műegyetem rkp. 3, H-1111 Budapest, Hungary
1 shared publication - Mohammad Dowajy
Department of Photogrammetry and Geoinformatics, Faculty of Civil Engineering, Budapest University of Technology and Economics, Műegyetem rkp. 3, H-1111 Budapest, Hungary
1 shared publication - Oybek Tukhtamishov
Department of Geodesy and Geoinformatics, Tashkent Institute of Irrigation and Agricultural Mechanization Engineers (TIIAME)—National Research University, 39 Koriy Niyoziy str., Tashkent 100000, Uzbekistan
1 shared publication
Latest Publications
- Open Access
- 4 August 2026
- Published by MDPI AG in Remote Sensing
- Vol. 18 (15) , 2571
A Comprehensive Machine Learning Approach for Crop Classification Using Multi-Sensor Satellite Datasets and Multiple Vegetation Indices
Accurate crop classification is essential for sustainable agriculture activities and food security studies. Recent advancements in remote sensing data acquisition and analysis techniques enable various solutions for cropland detection; however, reliable crop maps are still lacking in many heterogeneous semi-arid regions (e.g., Central Asia). Machine learning approaches address such challenges and distinguish different crop types using multiple datasets. The main aim of this study is to optimize crop classification outcomes by integrating multi-sensor datasets leveraging numerous vegetation indices through different machine learning models. Four datasets: Landsat-8 (DS-1), Sentinel-2 (DS-2), optical Sentinel-2 integrated with SAR Sentinel-1 (DS-3), and Sentinel-1 (DS-4) were used for the developed experiments. Five vegetation indices, NDVI, GNDVI, EVI, SAVI, and MSAVI, were derived using Sentinel-2 and Landsat-8 bands; in addition, NDRE was only obtained for Sentinel-2 exploiting the red edge band. Three input scenarios were considered for model training and image classification, featuring solely NDVI and its related bands; a set of vegetation indices and their associated bands for optical imagery; and VV, VH, and VV/VH ratio bands for SAR data. Five classifiers, Gradient Boosting Tree (GBT), Random Forest (RF), K-Nearest Neighbor (KNN), Classification and Regression Tree (CART), and Minimum Distance (MD), were employed to assess the machine learning quality for scene classification. Findings demonstrated that Sentinel-2 outperforms Landsat-8 images due to the higher spatial resolution and red edge bands. DS-3 consistently outperforms both DS-2 (optical-only) and DS-4 (SAR-only) across all classifiers, enhancing the overall accuracy up to 2.38% over the optical dataset, and up to 13.28% over the SAR data, demonstrating the added details on canopy spectral reflectance, structure and moisture content. Using multiple vegetation indices consistently improves performance over NDVI alone across DS-1, DS-2, and DS-3, with gains reaching up to 96.22% due to the complementary information captured by multi-index spectral sensitivity. The GBT and RF classifiers consistently achieved the highest classification performance, effectively combining multiple decision trees to capture complex nonlinear relationships and decision boundaries; meanwhile, the MD classifier exhibited the lowest accuracy due to its reliance solely on distances to class mean vectors. All in all, the presented approach offers a robust framework for crop classification supplemented with multiple data sources using different VI feature scenarios and variable machine learning tools for precise farming applications in semi-arid regions.
- 4 August 2026
- Published by MDPI AG in Remote Sensing
- Vol. 18 (15) , 2571
Open AccessA Comprehensive Machine Learning Approach for Crop Classification Using Multi-Sensor Satellite Datasets and Multiple Vegetation Indices
Accurate crop classification is essential for sustainable agriculture activities and food security studies. Recent advancements in remote sensing data acquisition and analysis techniques enable various solutions for cropland detection; however, reliable crop maps are still lacking in many heterogeneous semi-arid regions (e.g., Central Asia). Machine learning approaches address such challenges and distinguish different crop types using multiple datasets. The main aim of this study is to optimize crop classification outcomes by integrating multi-sensor datasets leveraging numerous vegetation indices through different machine learning models. Four datasets: Landsat-8 (DS-1), Sentinel-2 (DS-2), optical Sentinel-2 integrated with SAR Sentinel-1 (DS-3), and Sentinel-1 (DS-4) were used for the developed experiments. Five vegetation indices, NDVI, GNDVI, EVI, SAVI, and MSAVI, were derived using Sentinel-2 and Landsat-8 bands; in addition, NDRE was only obtained for Sentinel-2 exploiting the red edge band. Three input scenarios were considered for model training and image classification, featuring solely NDVI and its related bands; a set of vegetation indices and their associated bands for optical imagery; and VV, VH, and VV/VH ratio bands for SAR data. Five classifiers, Gradient Boosting Tree (GBT), Random Forest (RF), K-Nearest Neighbor (KNN), Classification and Regression Tree (CART), and Minimum Distance (MD), were employed to assess the machine learning quality for scene classification. Findings demonstrated that Sentinel-2 outperforms Landsat-8 images due to the higher spatial resolution and red edge bands. DS-3 consistently outperforms both DS-2 (optical-only) and DS-4 (SAR-only) across all classifiers, enhancing the overall accuracy up to 2.38% over the optical dataset, and up to 13.28% over the SAR data, demonstrating the added details on canopy spectral reflectance, structure and moisture content. Using multiple vegetation indices consistently improves performance over NDVI alone across DS-1, DS-2, and DS-3, with gains reaching up to 96.22% due to the complementary information captured by multi-index spectral sensitivity. The GBT and RF classifiers consistently achieved the highest classification performance, effectively combining multiple decision trees to capture complex nonlinear relationships and decision boundaries; meanwhile, the MD classifier exhibited the lowest accuracy due to its reliance solely on distances to class mean vectors. All in all, the presented approach offers a robust framework for crop classification supplemented with multiple data sources using different VI feature scenarios and variable machine learning tools for precise farming applications in semi-arid regions.
- Open Access
- 28 July 2025
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-G-2, 459-465
High-Quality Road Detection Using U-Net-Based Semantic Segmentation with High-Resolution Orthophotos and DSM Data in Urban Environments
Road detection and recognition from high-resolution geospatial data in urban environments is critical for numerous applications, including urban planning, navigation systems, and automated driving technologies. This study explores the potential of deep learning methodologies, specifically U-Net-based semantic segmentation, for high-quality road detection using open-access datasets. The input data consists of high-resolution digital orthophotos of a city region and corresponding Digital Surface Models (DSMs), allowing for a comprehensive analysis of two scenarios: (A) semantic segmentation using only imagery data and (B) segmentation utilizing both imagery and DSM data. Building on prior works by the authors, which include digital surface modelling and satellite image classification using U-Net and other neural network architectures, this research applies state-of-the-art techniques to leverage the spatial richness of orthophotos and the vertical information embedded in DSMs. Preliminary results indicate a significant improvement in road detection accuracy when integrating DSM data, highlighting the synergistic value of multi-source data in geospatial analysis. To validate the segmentation outputs, independent satellite imagery data are employed as a benchmark, enabling quantitative assessments of positional accuracy. The integrated orthophoto-DSM strategy achieved completeness, correctness, quality, and overall accuracy of 99.76%, 77.35%, 77.21%, and 95.95%, respectively, surpassing the solely orthophoto-based model, which achieved 73.76%, 51.10%, 42.52%, and 86.29% in these metrics. Experiments ensure that the proposed methodology achieves a robust delineation of road networks in urban environments, with scenario B outperforming scenario A. This research contributes to the growing body of literature on deep learning applications in photogrammetry and remote sensing, aligning with related studies on the semantic segmentation of urban features. The results underscore the importance of multimodal data fusion in geospatial analysis and its implications for enhancing road detection frameworks. Extracted road maps provide an accurate baseline for urban development planning and transportation management applications.
- 28 July 2025
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-G-2, 459-465
Open AccessHigh-Quality Road Detection Using U-Net-Based Semantic Segmentation with High-Resolution Orthophotos and DSM Data in Urban Environments
Road detection and recognition from high-resolution geospatial data in urban environments is critical for numerous applications, including urban planning, navigation systems, and automated driving technologies. This study explores the potential of deep learning methodologies, specifically U-Net-based semantic segmentation, for high-quality road detection using open-access datasets. The input data consists of high-resolution digital orthophotos of a city region and corresponding Digital Surface Models (DSMs), allowing for a comprehensive analysis of two scenarios: (A) semantic segmentation using only imagery data and (B) segmentation utilizing both imagery and DSM data. Building on prior works by the authors, which include digital surface modelling and satellite image classification using U-Net and other neural network architectures, this research applies state-of-the-art techniques to leverage the spatial richness of orthophotos and the vertical information embedded in DSMs. Preliminary results indicate a significant improvement in road detection accuracy when integrating DSM data, highlighting the synergistic value of multi-source data in geospatial analysis. To validate the segmentation outputs, independent satellite imagery data are employed as a benchmark, enabling quantitative assessments of positional accuracy. The integrated orthophoto-DSM strategy achieved completeness, correctness, quality, and overall accuracy of 99.76%, 77.35%, 77.21%, and 95.95%, respectively, surpassing the solely orthophoto-based model, which achieved 73.76%, 51.10%, 42.52%, and 86.29% in these metrics. Experiments ensure that the proposed methodology achieves a robust delineation of road networks in urban environments, with scenario B outperforming scenario A. This research contributes to the growing body of literature on deep learning applications in photogrammetry and remote sensing, aligning with related studies on the semantic segmentation of urban features. The results underscore the importance of multimodal data fusion in geospatial analysis and its implications for enhancing road detection frameworks. Extracted road maps provide an accurate baseline for urban development planning and transportation management applications.
- Open Access
- 28 July 2025
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-G-2, 389-394
Automatic Non-Urban Road Surface Point Extraction Based on Geometric Features Using Neural Networks and Raster Structure Approach
The automatic segmentation of road surface points from 3D point cloud data has recently gained significant attention. However, it remains challenging due to the variability of road characteristics and the complexity of surrounding environments, especially in non-urban areas. This study introduces a comprehensive methodology leveraging neural networks to segment road surface points from non-urban point cloud data, supporting autonomous driving applications. The proposed approach computes multiple geometric features of the point cloud at two resolutions, 0.2 and 0.4 meters, to enhance segmentation accuracy. The features are projected onto a regular grid and converted into a raster format where each pixel's value represents the averaged features of points within its space. The rasterized values serve as structured inputs for a feature-based Neural Network (NN), which classifies road pixels based on intensity, density, curvature, planarity, roughness, surface variation, and verticality properties. Classified road pixels are further refined through morphological operations, distinguishing main road and road border pixels. The created masks are then used to extract the corresponding point cloud data of each category. The same neural network model is applied to extract road points within the border point clouds, where a precise road surface point cloud is obtained by merging the inside-road and filtered border points. The proposed method was evaluated on a MLS-acquired road point cloud dataset, achieving high performance with average completeness, correctness, quality, and overall accuracy rates of 98.9%, 97.6%, 96.6%, and 98.2%, respectively. Its key advantage lies in the reduced computational requirements demand by operating on rasterized inputs rather than traditional raw point cloud data.
- 28 July 2025
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-G-2, 389-394
Open AccessAutomatic Non-Urban Road Surface Point Extraction Based on Geometric Features Using Neural Networks and Raster Structure Approach
The automatic segmentation of road surface points from 3D point cloud data has recently gained significant attention. However, it remains challenging due to the variability of road characteristics and the complexity of surrounding environments, especially in non-urban areas. This study introduces a comprehensive methodology leveraging neural networks to segment road surface points from non-urban point cloud data, supporting autonomous driving applications. The proposed approach computes multiple geometric features of the point cloud at two resolutions, 0.2 and 0.4 meters, to enhance segmentation accuracy. The features are projected onto a regular grid and converted into a raster format where each pixel's value represents the averaged features of points within its space. The rasterized values serve as structured inputs for a feature-based Neural Network (NN), which classifies road pixels based on intensity, density, curvature, planarity, roughness, surface variation, and verticality properties. Classified road pixels are further refined through morphological operations, distinguishing main road and road border pixels. The created masks are then used to extract the corresponding point cloud data of each category. The same neural network model is applied to extract road points within the border point clouds, where a precise road surface point cloud is obtained by merging the inside-road and filtered border points. The proposed method was evaluated on a MLS-acquired road point cloud dataset, achieving high performance with average completeness, correctness, quality, and overall accuracy rates of 98.9%, 97.6%, 96.6%, and 98.2%, respectively. Its key advantage lies in the reduced computational requirements demand by operating on rasterized inputs rather than traditional raw point cloud data.
- Open Access
- 10 July 2025
- Published by Copernicus GmbH in ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. X-G-2025, 245-252
VHR Multispectral Satellite Image Classification with Kolmogorov-Arnold Networks for Urban Applications
The Kolmogorov-Arnold Network (KAN) is a computational framework rooted in the Kolmogorov-Arnold representation theorem, which states that any continuous function of multiple variables can be expressed as a superposition of univariate functions. KAN leverages this theoretical foundation to decompose complex functions into simpler and lower-dimensional ones. Function decomposition makes KAN particularly suitable for challenging tasks in image classification, where separating data into distinct categories often requires approximating intricate and multi-dimensional boundaries. In practice, KAN, relying on function decomposition, offers an alternative to multilayer perceptron (MLP) architectures. By efficiently encoding nonlinear relationships among features, KAN demonstrates potential in multilayered data analysis tasks such as multi and hyper-spectral remotely sensed image classification. As a result, KAN finds significant applications in remote sensing image processing, particularly in Land Cover and Land Use (LCLU) mapping using Very High-Resolution (VHR) satellite imageries. Due to their fine detail, VHR images provide an exceptional basis for accurately distinguishing between various land cover types. When multispectral data is incorporated, KAN excels by leveraging its ability to model nonlinear relationships, allowing for highly accurate classifications. KAN's performance is validated using ground-truth data collected from field surveys or random reference points visually picked from images. Additionally, KAN is benchmarked against other methods like traditional Shallow Neural Networks (SNNs). Obtained KAN classification accuracy and computational efficiency are evaluated compared to SNN, highlighting its strengths in modelling complexity with optimized key model parameters. A Python-based implementation of KAN is developed for flexible integration into existing geospatial analysis workflows and highlighting its compatibility with cloud computing environments such as Google Colab. This integration enhances scalability, makes it practical for processing large-scale satellite datasets efficiently, and facilitates high-resolution mapping and reproducibility in environmental monitoring and urban applications. The reliability of KAN and the potential classification accuracy of different model architectures were verified. The KAN model with a 10-neuron mid-layer achieved an overall accuracy of 88.89%, outperforming the SNN results with a maximum accuracy of 87.84 for a model with 20 & 20-neuron hidden layers.
- 10 July 2025
- Published by Copernicus GmbH in ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. X-G-2025, 245-252
Open AccessVHR Multispectral Satellite Image Classification with Kolmogorov-Arnold Networks for Urban Applications
The Kolmogorov-Arnold Network (KAN) is a computational framework rooted in the Kolmogorov-Arnold representation theorem, which states that any continuous function of multiple variables can be expressed as a superposition of univariate functions. KAN leverages this theoretical foundation to decompose complex functions into simpler and lower-dimensional ones. Function decomposition makes KAN particularly suitable for challenging tasks in image classification, where separating data into distinct categories often requires approximating intricate and multi-dimensional boundaries. In practice, KAN, relying on function decomposition, offers an alternative to multilayer perceptron (MLP) architectures. By efficiently encoding nonlinear relationships among features, KAN demonstrates potential in multilayered data analysis tasks such as multi and hyper-spectral remotely sensed image classification. As a result, KAN finds significant applications in remote sensing image processing, particularly in Land Cover and Land Use (LCLU) mapping using Very High-Resolution (VHR) satellite imageries. Due to their fine detail, VHR images provide an exceptional basis for accurately distinguishing between various land cover types. When multispectral data is incorporated, KAN excels by leveraging its ability to model nonlinear relationships, allowing for highly accurate classifications. KAN's performance is validated using ground-truth data collected from field surveys or random reference points visually picked from images. Additionally, KAN is benchmarked against other methods like traditional Shallow Neural Networks (SNNs). Obtained KAN classification accuracy and computational efficiency are evaluated compared to SNN, highlighting its strengths in modelling complexity with optimized key model parameters. A Python-based implementation of KAN is developed for flexible integration into existing geospatial analysis workflows and highlighting its compatibility with cloud computing environments such as Google Colab. This integration enhances scalability, makes it practical for processing large-scale satellite datasets efficiently, and facilitates high-resolution mapping and reproducibility in environmental monitoring and urban applications. The reliability of KAN and the potential classification accuracy of different model architectures were verified. The KAN model with a 10-neuron mid-layer achieved an overall accuracy of 88.89%, outperforming the SNN results with a maximum accuracy of 87.84 for a model with 20 & 20-neuron hidden layers.