Search Results (23)

The advancement of artificial intelligence (AI) in remote sensing (RS) increasingly depends on datasets that offer rich and structured supervision beyond traditional scene-level labels. Although existing benchmarks for aerial scene classification have facilitated progress in this area, their reliance on single-class annotations restricts their application to more flexible, interpretable and generalisable learning frameworks. In this study, we introduce WHU-RS19 ABZSL: an attribute-based extension of the widely adopted WHU-RS19 dataset. This new version comprises 1005 high-resolution aerial images across 19 scene categories, each annotated with a vector of 38 features. These cover objects (e.g., roads and trees), geometric patterns (e.g., lines and curves) and dominant colours (e.g., green and blue), and are defined through expert-guided annotation protocols. To demonstrate the value of the dataset, we conduct baseline experiments using deep learning models that had been adapted for multi-label classification—ResNet18, VGG16, InceptionV3, EfficientNet and ViT-B/16—designed to capture the semantic complexity characteristic of real-world aerial scenes. The results, which are measured in terms of macro F1-score, range from 0.7385 for ResNet18 to 0.7608 for EfficientNet-B0. In particular, EfficientNet-B0 and ViT-B/16 are the top performers in terms of the overall macro F1-score and consistency across attributes, while all models show a consistent decline in performance for infrequent or visually ambiguous categories. This confirms that it is feasible to accurately predict semantic attributes in complex scenes. By enriching a standard benchmark with detailed, image-level semantic supervision, WHU-RS19 ABZSL supports a variety of downstream applications, including multi-label classification, explainable AI, semantic retrieval, and attribute-based ZSL. It thus provides a reusable, compact resource for advancing the semantic understanding of remote sensing and multimodal AI. Full article

Rapidly evolving surveying and monitoring methods are leading the transition toward more efficient, data-driven forest management practices. Recent research highlights the potential of advanced remote sensing platforms to support “smart” forestry, enabling precise, timely, and cost-effective assessments which inform multi-function management methods and specialized silvicultural practices for each forest type, composition, and structure. We created a digital replica of a marteloscope, which is a forestry tool to practice silvicultural simulations for technicians and students. The selected stand is an official marteloscope included in the Integrate+ Network project coordinated by the European Forest Institute (EFI). We established a framework for data collection and processing to achieve an accurate digital replica, using a mobile laser scanner (MLS) in a European beech (Fagus sylvatica L.) forest stand. We extracted the main structural forest parameters (diameter at breast height (DBH) and total height (TH)), using the 3DFin software and we graphically returned the obtained digital replica with the CloudCompare software. We compared the MLS-derived values of DBH (1087 trees) and TH (50 trees) with those from a traditional field survey and obtained a root mean square deviation (RMSD) of 2.38 cm for DBH and 2.42 m for TH. The digital marteloscope can help to visualize and assess the effects of selective thinning options on forest structure. The implementation of these virtual reality or augmented reality applications is a useful step toward smarter forestry and could be further improved. Full article

In the rapidly evolving field of remote sensing, Deep Learning (DL) techniques have become pivotal in interpreting and processing complex datasets. However, the increasing reliance on these algorithms necessitates a robust ethical framework to evaluate their trustworthiness. This paper introduces a comprehensive ethical framework designed to assess and quantify the trustworthiness of DL techniques in the context of remote sensing. We first define trustworthiness in DL as a multidimensional construct encompassing accuracy, reliability, transparency and explainability, fairness, and accountability. Our framework then operationalizes these dimensions through a set of quantifiable metrics, allowing for the systematic evaluation of DL models. To illustrate the applicability of our framework, we selected an existing case study in remote sensing, wherein we apply our ethical assessment to a DL model used for classification. Our results demonstrate the model’s performance across different trustworthiness metrics, highlighting areas for ethical improvement. This paper not only contributes a novel framework for ethical analysis in the field of DL, but also provides a practical tool for developers and practitioners in remote sensing to ensure the responsible deployment of DL technologies. Through a dual approach that combines top-down international standards with bottom-up, context-specific considerations, our framework serves as a practical tool for ensuring responsible AI applications in remote sensing. Its application through a case study highlights its potential to influence policy-making and guide ethical AI development in this domain. Full article

Due to ever-accelerating urbanization in recent decades, exploring the contributions of trees in mitigating atmospheric carbon in urban areas has become one of the paramount concerns. Remote sensing-based approaches have been primarily implemented to estimate the tree-stand atmospheric carbon stock (CS) for the trees in parks and streets. However, a convenient yet high-accuracy computation methodology is hardly available. This study introduces an approach that has been tested for a small urban area. A data fusion approach based on a three-dimensional (3D) computation methodology was applied to calibrate the individual tree CS. This photogrammetry-based technique employed an unmanned aerial vehicle (UAV) and spherical image data to compute the total height (H) and diameter at breast height (DBH) for each tree, consequently estimating the tree-stand CS. A regression analysis was conducted to compare the results with the ones obtained with high-cost laser scanner data. Our study demonstrates the applicability of this method, highlighting its advantages even for large city areas in contrast to other approaches that are often more expensive. This approach could serve as an efficient tool for assisting urban planners in ensuring the proper utilization of the available green space, especially in a complex urban environment. Full article

The seamless integration of indoor and outdoor positioning has gained considerable attention due to its practical implications in various fields. This paper presents an innovative approach aimed at detecting and delineating outdoor, indoor, and transition areas using a time series analysis of Global Navigation Satellite System (GNSS) error statistics. By leveraging this contextual understanding, the decision-making process between GNSS-based and Visual-Inertial Odometry (VIO) for trajectory estimation is refined, enabling a more robust and accurate positioning. The methodology involves three key steps: proposing the division of our context environment into a set of areas (indoor, outdoor, and transition), exploring two methodologies for the classification of space based on a time series of GNSS error statistics, and refining the trajectory estimation strategy based on contextual knowledge. Real data across diverse scenarios validate the approach, yielding trajectory estimations with accuracy consistently below 10 m. Full article

Geoinformation and aerial data collection are essential during post-earthquake emergency response. This research focuses on the long-lasting spatial impacts of temporary solutions, which have persisted in regions of Central Italy affected by catastrophic seismic events over the past 25 years, significantly and permanently altering their landscapes. The paper analyses the role of geomatic and photogrammetric tools in documenting the emergency process and projects in post-disaster phases. An Atlas of Temporary Architectures is proposed, which defines a common semantic and geometric codification for mapping temporary housing from territorial to urban and building scales. The paper presents an implementation of attribute specification in existing official cartographic data, including geometric entities in a 3D GIS data model platform for documenting and digitalising these provisional contexts. To achieve this platform, UAV point clouds are integrated with non-metric data to ensure a complete description in a multiscalar approach. Accurate topographic modifications can be captured by extracting very high-resolution orthophotos and elevation models (DSM and DTM). The results have been validated in Visso (Macerata), a small historical mountain village in Central Italy which was heavily damaged by the seismic events of 2016/2017. The integrated approach overcomes the existing gaps and emphasizes the importance of managing heterogeneous geospatial emergency data for classification purposes. It also highlights the need to enhance an interoperable knowledge base method for post-disaster temporary responses. By combining geomatic tools with architectural studies, these visualization techniques can support national and local organizations responsible for post-earthquake management through a 3D modelling method to aid future transformations or interventions following other natural disasters. Full article

In recent years, advancements in remote and proximal sensing technology have driven innovation in environmental and land surveys. The integration of various geomatics devices, such as reflex and UAVs equipped with RGB cameras and mobile laser scanners (MLS), allows detailed and precise surveys of monumental trees. With these data fusion method, we reconstructed three monumental 3D tree models, allowing the computation of tree metric variables such as diameter at breast height (DBH), total height (TH), crown basal area (CBA), crown volume (CV) and wood volume (WV), even providing information on the tree shape and its overall conditions. We processed the point clouds in software such as CloudCompare, 3D Forest, R and MATLAB, whereas the photogrammetric processing was conducted with Agisoft Metashape. Three-dimensional tree models enhance accessibility to the data and allow for a wide range of potential applications, including the development of a tree information model (TIM), providing detailed data for monitoring tree health, growth, biomass and carbon sequestration. The encouraging results provide a basis for extending the virtualization of these monumental trees to a larger scale for conservation and monitoring. Full article

The scarce knowledge and documentation of Underground Built Heritage (UBH) assets frequently limit their full exploitation and valorization. The aim of this work is to reflect on the techniques, functions, and technical features of a specific case study in a very broad context that can, however, be a building block for the understanding, preservation, and reuse of architectural and engineering values that represent a fundamental trace of the history of a society. Therefore, to fill these knowledge gaps, it was constructed a 3D GIS model, multi-scale, and interoperable database, capable of management, promotion, and valorization of UBH. The case study focuses on the old water supply system of the city of Lisbon, as UBH site, with galleries and cisterns that are points of connection with the urban environment above. For the creation of 3D models of the structure under investigation, it was decided to carry out a survey with Mobile Mapping System as a first step, which allowed the construction of a dense point cloud useful to build 3D models of individual objects. Finally, the 3D models were imported into the 3D GIS environment and multi-information could be linked for each previously identified element for greater knowledge sharing. This research has demonstrated how geomatic techniques can be effectively used in conjunction with the information management systems of GIS to explore this “hidden” heritage and has highlighted the limitations and problems of 3D digitization of the UBH. The results obtained offer the possibility of extending and adapting the methodology to different application contexts and the possibility of customizing the data representation. Full article

The construction technique of raw earth, which has always been in use in most of the world, has left large monuments or architectural complexes to cultural heritage that need special attention due to the notable vulnerability of the material. A convenient way to deal this threat, besides physical intervention, is by using an information system, such as HBIM (Heritage Building Information Modeling), as a tool for damage assessment and conservation planning. This paper reports on its application in an archaeological setting, in particular, on the Huaca Arco Iris, a religious building of the old city of Chan Chan (Peru), the largest monumental complex in mud on the American continent. The study is part of the bilateral international project between the Consiglio Nazionale delle Ricerche (CNR) and the Consejo Nacional de Ciencia, Tecnología e Innovación Tecnológica (CONCYTEC) in the use of HBIM for the prediction of possible natural or anthropogenic damages to buildings in raw mud. Exploiting the data coming from the direct and indirect analyses, a dedicated ontology is built to guide the management of these data within the information system. The creation of an HBIM system for the archaeological domain, based on the trinomial data–information–knowledge, is presented and validated. Following this approach, a customizable HBIM has been created with the 3D model of the spatial entities of the Huaca. As a result, the semantic relationship of an external wall, taken as the benchmark test of our experiment, with the contained bas-relief and the conservation cover is tested. Full article

Historical villages represent a highly vulnerable cultural heritage; their preservation can be ensured thanks to technological innovations in the field of geomatics and information systems. Among these, Geographical Information Systems (GISs) allow exploiting heterogeneous data for efficient vulnerability assessment, in terms of both time and usability. Geometric attributes, which currently are mainly inferred by visual inspections, can be extrapolated from data obtained by geomatic technologies. Furthermore, the integration with non-metric data ensures a more complete description of the post-seismic risk thematic mapping. In this paper, a high-performance information system for small urban realities, such as historical villages, is described, starting from the 3D survey obtained through the integrated management of recent innovative geomatic sensors, such as Unmanned Aerial Vehicles (UAVs), Terrestrial Laser Scanners (TLSs), and 360º images. The results show that the proposed strategy of the automatic extraction of the parameters from the GIS can be generalized to other case studies, thus representing a straightforward method to enhance the decision-making of public administrations. Moreover, this work confirms the importance of managing heterogeneous geospatial data to speed up the vulnerability assessment process. The final result, in fact, is an information system that can be used for every village where data have been acquired in a similar way. This information could be used in the field by means of a GIS app that allows updating the geospatial database, improving the work of technicians. This approach was validated in Gabbiano(Pieve Torina), a village in Central Italy affected by earthquakes in 2016 and 2017. Full article

This paper introduces a new method for determining the shape similarity of complex three-dimensional (3D) mesh structures based on extracting a vector of important vertices, ordered according to a matrix of their most important geometrical and topological features. The correlation of ordered matrix vectors is combined with perceptual definition of salient regions in order to aid detection, distinguishing, measurement and restoration of real degradation and digitization errors. The case study is the digital 3D structure of the Camino Degli Angeli, in the Urbino’s Ducal Palace, acquired by the structure from motion (SfM) technique. In order to obtain an accurate, featured representation of the matching shape, the strong mesh processing computations are performed over the mesh surface while preserving real shape and geometric structure. In addition to perceptually based feature ranking, the new theoretical approach for ranking the evaluation criteria by employing neural networks (NNs) has been proposed to reduce the probability of deleting shape points, subject to optimization. Numerical analysis and simulations in combination with the developed virtual reality (VR) application serve as an assurance to restoration specialists providing visual and feature-based comparison of damaged parts with correct similar examples. The procedure also distinguishes mesh irregularities resulting from the photogrammetry process. Full article

Among the existing Cultural Heritage settings, Underground Built Heritage (UBH) represents a peculiar case. The scarce or lack of knowledge and documentation of these spaces frequently limits their proper management, exploitation, and valorization. When mapping these environments for documentation purposes, the primary need is to achieve a complete, reliable, and adequate representation of the built spaces and their geometry. Terrestrial laser scanners were widely employed for this task, although the procedure is generally time-consuming and often lacks color information. Mobile Mapping Systems (MMSs) are nowadays fascinating and promising technologies for mapping underground structures, speeding up acquisition times. In this paper, mapping experiences (with two commercial tools and an in-house prototype) in UBH settings are presented, testing the different handheld mobile solutions to guarantee an accurate and reliable 3D digitization. Tests were performed in the selected case study of Camerano Caves (Italy), characterized by volumetric complexity, poor lighting conditions, and difficult accessibility. The aim of this research activity is not only to show the differences in the technological instruments used for 3D surveying, but rather to argue over the pros and cons of the systems, providing the community with best practices and rules for 3D data collection with handheld mobile systems. The experiments deliver promising results when compared with TLS data. Full article

Renewable energy sources will represent the only alternative to limit fossil fuel usage and pollution. For this reason, photovoltaic (PV) power plants represent one of the main systems adopted to produce clean energy. Monitoring the state of health of a system is fundamental. However, these techniques are time demanding, cause stops to the energy generation, and often require laboratory instrumentation, thus being not cost-effective for frequent inspections. Moreover, PV plants are often located in inaccessible places, making any intervention dangerous. In this paper, we propose solAIr, an artificial intelligence system based on deep learning for anomaly cells detection in photovoltaic images obtained from unmanned aerial vehicles equipped with a thermal infrared sensor. The proposed anomaly cells detection system is based on the mask region-based convolutional neural network (Mask R-CNN) architecture, adopted because it simultaneously performs object detection and instance segmentation, making it useful for the automated inspection task. The proposed system is trained and evaluated on the photovoltaic thermal images dataset, a publicly available dataset collected for this work. Furthermore, the performances of three state-of-art deep neural networks, (DNNs) including UNet, FPNet and LinkNet, are compared and evaluated. Results show the effectiveness and the suitability of the proposed approach in terms of intersection over union (IoU) and the Dice coefficient. Full article

This contribution describes the methodology applied to evaluate the suitability of a Long-Range Mobile Mapping System to be integrated with other techniques that are currently used in a large and complex landslide deformation monitoring project carried out in Cortes de Pallás, in Valencia (Spain). Periodical geodetic surveys provide a reference frame realized by 10 pillars and 15 additional check points placed in specific points of interest, all with millimetric accuracy. The combined use of Close-Range Photogrammetry provides a well-controlled 3D model with 1–3 cm accuracy, making the area ideal for testing new technologies. Since some zones of interest are usually obstructed by construction, trees, or lamp posts, a possible solution might be the supplementary use of dynamic scanning instruments with the mobile mapping solution Kaarta Stencil 2 to collect the missing data. However, the reliability of this technology has to be assessed and validated before being integrated into the existing 3D models in the well-controlled area of Cortes de Pallás. The results of the experiment show that the accuracy achieved are compatible with those obtained from Close-Range Photogrammetry and can also be safely used to supplement image-based information for monitoring with 3–8 cm overall accuracy. Full article

In recent years semantic segmentation of 3D point clouds has been an argument that involves different fields of application. Cultural heritage scenarios have become the subject of this study mainly thanks to the development of photogrammetry and laser scanning techniques. Classification algorithms based on machine and deep learning methods allow to process huge amounts of data as 3D point clouds. In this context, the aim of this paper is to make a comparison between machine and deep learning methods for large 3D cultural heritage classification. Then, considering the best performances of both techniques, it proposes an architecture named DGCNN-Mod+3Dfeat that combines the positive aspects and advantages of these two methodologies for semantic segmentation of cultural heritage point clouds. To demonstrate the validity of our idea, several experiments from the ArCH benchmark are reported and commented. Full article