High-resolution Aerial Imagery Research Articles

Combining dual attention mechanism and efficient feature aggregation for road and vehicle segmentation from UAV imagery

Unmanned aerial vehicles (UAVs) have gained significant popularity in recent years due to their ability to capture high-resolution aerial imagery for various applications, including traffic monitoring, urban planning, and disaster management. Accurate road and vehicle segmentation from UAV imagery plays a crucial role in these applications. In this paper, we propose a novel approach combining dual attention mechanisms and efficient multi-layer feature aggregation to enhance the performance of road and vehicle segmentation from UAV imagery. Our approach integrates a spatial attention mechanism and a channel-wise attention mechanism to enable the model to selectively focus on relevant features for segmentation tasks. In conjunction with these attention mechanisms, we introduce an efficient multi-layer feature aggregation method that synthesizes and integrates multi-scale features at different levels of the network, resulting in a more robust and informative feature representation. Our proposed method is evaluated on the UAVid semantic segmentation dataset, showcasing its exceptional performance in comparison to renowned approaches such as U-Net, DeepLabv3+, and SegNet. The experimental results affirm that our approach surpasses these state-of-the-art methods in terms of segmentation accuracy.

National-scale nighttime high-temperature anomalies from Landsat-8 OLI images

Identifying and characterizing the distribution, abundance, and structural composition of national high-temperature anomalies (HTAs) are essential for mitigating natural disasters, regulating industrial emissions, and assessing societal impacts. The Landsat-8 Operational Land Imager (OLI) provides a unique capability for monitoring nighttime HTAs (N-HTAs) at high spatial resolutions without contamination from reflected solar radiation. However, the extraction of N-HTAs from nighttime OLI imagery poses significant challenges due to the interference of bright up-light, especially for smaller and cooler hotspots, and the absence of spectral information on backgrounds complicates the classification of N-HTAs at night. Here, we explored the background noise characteristics of Bands 2–7 in nighttime OLI imagery and developed an object-oriented approach for the enhanced detection of small and cool N-HTAs. Furthermore, we introduced a deep learning model that leverages high-resolution aerial imagery and land cover data to categorize N-HTAs into industrial sources, living sources, vegetation fires, and others. Applying these methodologies to nighttime OLI imagery from 2013 to 2020 allowed us to map and classify N-HTAs across the contiguous United States (CONUS), facilitating the creation of a detailed spatiotemporal inventory of industrial N-HTAs. Our analysis revealed that N-HTAs from living sources dominate the total N-HTAs, with a pronounced holiday effect, and gas flaring activities in oil and gas development areas account for >85 % of industrial N-HTAs. The outcomes of this national-scale investigation underscore the significant potential and practical utility of nighttime OLI imagery for N-HTA monitoring and advocate for expanding the spatial and temporal coverage of nighttime OLI observations globally.

Extraction of Water Bodies from High-Resolution Aerial and Satellite Images Using Visual Foundation Models

Water, indispensable for life and central to ecosystems, human activities, and climate dynamics, requires rapid and accurate monitoring. This is vital for sustaining ecosystems, enhancing human welfare, and effectively managing land, water, and biodiversity on both the local and global level. In the rapidly evolving domain of remote sensing and deep learning, this study focuses on water body extraction and classification through the use of recent deep learning models of visual foundation models (VFMs). Specifically, the Segment Anything Model (SAM) and Contrastive Language-Image Pre-training (CLIP) models have shown promise in semantic segmentation, dataset creation, change detection, and instance segmentation tasks. A novel two-step approach involving segmenting images via the Automatic Mask Generator method of the SAM and the zero-shot classification of segments using CLIP is proposed, and its effectiveness is tested on water body extraction problems. The proposed methodology was applied to both remote sensing imagery acquired from LANDSAT 8 OLI and very high-resolution aerial imagery. Results revealed that the proposed methodology accurately delineated water bodies across complex environmental conditions, achieving a mean intersection over union (IoU) of 94.41% and an F1 score of 96.97% for satellite imagery. Similarly, for the aerial imagery dataset, the proposed methodology achieved a mean IoU of 90.83% and an F1 score exceeding 94.56%. The high accuracy achieved in selecting segments predominantly classified as water highlights the effectiveness of the proposed model in intricate environmental image analysis.

Using Deep Learning and Advanced Image Processing for the Automated Estimation of Tornado-Induced Treefall

Each year, numerous tornadoes occur in forested regions of the United States. Due to the substantial number of fallen trees and accessibility issues, many of these tornadoes remain poorly documented and evaluated. The process of documenting tree damage to assess tornado intensity is known as the treefall method, an established and reliable technique for estimating near-surface wind speed. Consequently, the demand for documenting fallen trees has increased in recent years. However, the treefall method proves to be extremely expensive and time-consuming, requiring a laborious assessment of each treefall instance. This research proposes a novel approach to evaluating treefall in large, forested regions using deep learning-based automated detection and advanced image processing techniques. The developed treefall method relies on high-resolution aerial imagery from a damaged forest and involves three main steps: (1) instance segmentation detection, (2) estimating tree taper and predicting fallen tree directions, and (3) obtaining subsampled treefall vector results indicating the predominant flow direction in geospatial coordinates. To demonstrate the method’s effectiveness, the algorithm was applied to a tornado track rated EF-4, which occurred on 10 December 2021, cutting through the Land Between the Lakes National Recreation Area in Kentucky. Upon observation of the predicted results, the model is demonstrated to accurately predict the predominant treefall angles. This deep-learning-based treefall algorithm has the potential to speed up data processing and facilitate the application of treefall methods in tornado evaluation.

Sub-meter tree height mapping of California using aerial images and LiDAR-informed U-Net model

Tree canopy height is one of the most important indicators of forest biomass, productivity, and ecosystem structure, but it is challenging to measure accurately from the ground and from space. Here, we used a U-Net model adapted for regression to map the canopy height of all trees in the state of California with very high-resolution aerial imagery 0.6 m from the USDA-NAIP program. The U-Net model was trained using canopy height models computed from aerial LiDAR data as a reference, along with corresponding RGB-NIR NAIP images collected in 2020. We evaluated the performance of the deep-learning model using 42 independent 1 km2 areas across various forest types and landscape variations in California. Our predictions of tree heights exhibited a mean error of 2.9 m and showed relatively low systematic bias across the entire range of tree heights present in California. In 2020, trees taller than 5 m covered ∼ 19.3% of California. Our model successfully estimated canopy heights up to 50 m without saturation, outperforming existing canopy height products from global models. The approach we used allowed for the reconstruction of the three-dimensional structure of individual trees as observed from nadir-looking optical airborne imagery, suggesting a relatively robust estimation and mapping capability, even in the presence of image distortion. These findings demonstrate the potential of large-scale mapping and monitoring of tree height, as well as potential biomass estimation, using NAIP imagery.

Deep convolutional encoder–decoder networks based on ensemble learning for semantic segmentation of high-resolution aerial imagery

Deep convolutional encoder–decoder networks based on ensemble learning for semantic segmentation of high-resolution aerial imagery

Channel morphological change monitoring using high-resolution LiDAR-derived DEM and multi-temporal imageries

Natural processes and human activities both cause morphological changes in channels. Remote sensing products are often used to assess planform changes, but they tend to overlook vertical changes. However, considering both planform and vertical changes is crucial for a comprehensive evaluation of morphological changes. Using spatiotemporal aerial imagery and topographic data, remote sensing plays a vital role in evaluating channel morphological changes and flood-carrying capacity. This study aimed to investigate the morphological changes of a creek in an urban catchment using very high-resolution remote sensing products. In this study, we developed a new framework for investigating overall channel morphology change by employing very high-resolution aerial imagery and a LiDAR-derived digital elevation model (DEM). By digitizing channel boundaries using ArcGIS Pro 3.0, and analyzing various morphological parameters, erosion, and deposition patterns, we examined the impact of urban expansion and infrastructure development on channel adjustments. Channel adjustments have been performed in the case study catchment (Dry Creek, South Australia, Australia) due to urban expansion and development of infrastructure in the downstream reaches. Our findings revealed a significant southwest shift in the planform of the channel, with a maximum shift of 478 m and an average shift of 217 m since 1998. This alteration resulted in an increase in the sinuosity index reaching 1.2. Over the period from 2018 to 2022, the channel experienced a net deposition depth of 3.4 cm to 3.6 cm in downstream reaches. The annual deposition volume in the downstream reaches was 1963 m3, necessitating regular desilting to prevent channel capacity loss and flooding in the surrounding environment. This study also highlights the incremental growth of riparian vegetation within the channel, which affects surface roughness, channel slope, and carrying capacity. These findings provide a valuable baseline for future investigations into stream channel morphology changes and emphasize the importance of implementing appropriate measures such as desilting and vegetation management to mitigate deposition levels, reduce flood risks, and enhance the overall health and functionality of Dry Creek. The framework used in this study can be applied to other case studies employing reliable and high-resolution remote sensing data products.

An Attentive Hough Transform Module for Building Extraction From High Resolution Aerial Imagery

An Attentive Hough Transform Module for Building Extraction From High Resolution Aerial Imagery

Classified Identification and Estimation of behind-the-Meter Distributed Photovoltaic Panels Using High-Resolution Aerial Imagery

Classified Identification and Estimation of behind-the-Meter Distributed Photovoltaic Panels Using High-Resolution Aerial Imagery

Automated Large-Scale Tornado Treefall Detection and Directional Analysis Using Machine Learning

Abstract In many regions of the world, tornadoes travel through forested areas with low population densities, making downed trees the only observable damage indicator. Current methods in the EF scale for analyzing tree damage may not reflect the true intensity of some tornadoes. However, new methods have been developed that use the number of trees downed or treefall directions from high-resolution aerial imagery to provide an estimate of maximum wind speed. Treefall Identification and Direction Analysis (TrIDA) maps are used to identify areas of treefall damage and treefall directions along the damage path. Currently, TrIDA maps are generated manually, but this is labor-intensive, often taking several days or weeks. To solve this, this paper describes a machine learning– and image-processing-based model that automatically extracts fallen trees from large-scale aerial imagery, assesses their fall directions, and produces an area-averaged treefall vector map with minimal initial human interaction. The automated model achieves a median tree direction difference of 13.3° when compared to the manual tree directions from the Alonsa, Manitoba, tornado, demonstrating the viability of the automated model compared to manual assessment. Overall, the automated production of treefall vector maps from large-scale aerial imagery significantly speeds up and reduces the labor required to create a Treefall Identification and Direction Analysis map from a matter of days or weeks to a matter of hours. Significance Statement The automation of treefall detection and direction is significant to the analyses of tornado paths and intensities. Previously, it would have taken a researcher multiple days to weeks to manually count and assess the directions of fallen trees in large-scale aerial photography of tornado damage. Through automation, analysis takes a matter of hours, with minimal initial human interaction. Tornado researchers will be able to use this automated process to help analyze and assess tornadoes and their enhanced Fujita–scale rating around the world.

ON THE ASSESSMENT OF INSTANCE SEGMENTATION FOR THE AUTOMATIC DETECTION OF SPECIFIC CONSTRUCTIONS FROM VERY HIGH RESOLUTION AIRBORNE IMAGERY

Abstract. During the High Modernism period spanning from approximately 1914 to 1970, the manufacturing of steel-constructed system halls witnessed a significant surge to accommodate the growing demand across various sectors such as industry, commerce, and agriculture. Surprisingly, these specific types of buildings have been largely overlooked in the realm of construction history research, resulting in a dearth of knowledge regarding their construction methods, distribution patterns, and contextual significance for assessing their historical value. This study aims to address this gap by exploring the potential of instance segmentation methods for the automated detection of system halls using high-resolution aerial imagery. To achieve this objective, state-of-the-art deep learning models are evaluated in terms of their ability to localize and delineate system halls accurately. Our experiments reveal that Mask R-CNN yields the most accurate results both quantitatively and qualitatively, closely followed by Cascade Mask R-CNN. However, it is important to note that multi-scale methods may introduce false positives since system halls possess distinct geometric dimensions that necessitate careful consideration during the detection process.

Evaluating the effects of texture features on Pinus sylvestris classification using high-resolution aerial imagery

Pinus sylvestris, also known as Scots pine, is the most widespread type of pine tree found around the world. Texture features can recognize patterns in an image's texture and so distinguish different species with similar spectral properties. They also record information about the spatial layout of pixels inside an image. Nonetheless, previous research has not addressed the effects of textural variables on Scots pine classification using high-resolution aerial images. In the context of this investigation, we used TreeSatAI, a benchmark dataset with categorized high-resolution aerial imagery, to conduct a comparative analysis, and we examined a comprehensive set of 93 texture features (radiomics) by generating several parameters such as First Order, Gray Level Co-occurrence Matrix (GLCM), Gray Level Dependence Matrix (GLDM), Gray Level Size Zone Matrix (GLSZM), and Neighboring Gray Tone Difference Matrix (NGTDM). We proceeded to train machine learning models utilizing the state-of-the-art algorithms such as XGBoost, LightGBM, and Gradient Boosting Machines (GBM). Upon conducting a more comprehensive analysis, it has come to light that the GLDM and GLRLM features have exhibited greater significance in the realm of satellite image processing research, specifically pertaining to the precise classification of Scots pine. This finding challenges the conventional reliance on GLCM features, which have traditionally been favored in similar studies. This study is remarkable due to its emphasis on the efficacy of employing high-resolution aerial imagery and texture features within the realm of satellite image processing applications.

High-resolution forest canopy cover estimation in ecodiverse landscape using machine learning and Google Earth Engine: Validity and reliability assessment

The Forest Canopy Cover (FCC) is an important factor in the health and functioning of forests, as it plays a role in ecosystem services such as carbon sequestration, wildlife habitat, and water regulation. Techniques for accurately and efficiently mapping and extracting FCC information are evolving quickly, and it is of interest to evaluate their validity and reliability. The primary objectives of this study are to: 1) develop a large-scale FCC dataset with a spatial resolution of 1-m, 2) assess the spatial distribution of the FCC at a regional scale, and 3) examine the discrepancies in the FCC areas within the existing data from the Tree Canopy Cover (TCC) percent data by Hansen et al. (2013) and U.S. Forest Service TCC products at different spatial scales within the state of Arkansas (i.e., county and city levels). The FCC dataset was generated using high-resolution aerial imagery and a random forest (RF) machine learning algorithm, which were processed and analyzed using the Google Earth Engine cloud computing platform. The produced FCC dataset was validated using one-third of the observations (i.e., reference locations) obtained from the TCC percent data by Hansen et al. (2013) dataset and the 0.6-m spatial resolution National Agriculture Imagery Program (NAIP) aerial imagery. The results showed that the dataset successfully identified the FCC at a 1-m resolution in the study area, with an overall accuracy ranging between 83.31% and 94.35% per county. The spatial comparison results between the produced FCC dataset and both the Hansen et al. (2013) and USFS products also indicate a strong positive correlation, with the R2 values ranging between 0.94 and 0.98 for the county and the city levels. This dataset provides valuable information for monitoring, forecasting, and managing forest resources in Arkansas and other regions. Furthermore, the results of this study provide evidence that the use of machine learning and cloud computing technologies can produce high-resolution forest cover datasets and can be applied to other areas around the world.

Georectified polygon database of ground-mounted large-scale solar photovoltaic sites in the United States

Over 4,400 large-scale solar photovoltaic (LSPV) facilities operate in the United States as of December 2021, representing more than 60 gigawatts of electric energy capacity. Of these, over 3,900 are ground-mounted LSPV facilities with capacities of 1 megawatt direct current (MWdc) or more. Ground-mounted LSPV installations continue increasing, with more than 400 projects appearing online in 2021 alone; however, a comprehensive, publicly available georectified dataset including spatial footprints of these facilities is lacking. The United States Large-Scale Solar Photovoltaic Database (USPVDB) was developed to fill this gap. Using US Energy Information Administration (EIA) data, locations of 3,699 LSPV facilities were verified using high-resolution aerial imagery, polygons were digitized around panel arrays, and attributes were appended. Quality assurance and control were achieved via team peer review and comparison to other US PV datasets. Data are publicly available via an interactive web application and multiple downloadable formats, including: comma-separated value (CSV), application programming interface (API), and GIS shapefile and GeoJSON.

Cumulative effects of multiple stressors impact an endangered seagrass population and fish communities

Coastal ecosystems are becoming increasingly threatened by human activities and there is growing appreciation that management must consider the impacts of multiple stressors. Cumulative effects assessments (CEAs) have become a popular tool for identifying the distribution and intensity of multiple human stressors in coastal ecosystems. Few studies, however, have demonstrated strong correlations between CEAs and change in ecosystem condition, questioning its management use. Here, we apply a CEA to the endangered seagrass Posidonia australis in Pittwater, NSW, Australia, using spatial data on known stressors to seagrass related to foreshore development, water quality, vessel traffic and fishing. We tested how well cumulative effects scores explained changes in P. australis extent measured between 2005 and 2019 using high-resolution aerial imagery. A negative correlation between P. australis and estimated cumulative effects scores was observed (R2 = 22 %), and we identified a threshold of cumulative effects above which losses of P. australis became more likely. Using baited remote underwater video, we surveyed fishes over P. australis and non-vegetated sediments to infer and quantify how impacts of cumulative effects to P. australis extent would flow on to fish assemblages. P. australis contained a distinct assemblage of fish, and on non-vegetated sediments the abundance of sparids, which are of importance to fisheries, increased with closer proximity to P. australis. Our results demonstrate the negative impact of multiple stressors on P. australis and the consequences for fish biodiversity and fisheries production across much of the estuary. Management actions aimed at reducing or limiting cumulative effects to low and moderate levels will help conserve P. australis and its associated fish biodiversity and productivity.

Investigating the efficiency and capabilities of UAVs and Convolutional Neural Networks in the field of remote sensing as a land classification tool

The study aimed to determine the efficacy and capabilities of using high-resolution aerial imagery and a convolutional neural network (CNN) to identify plant species and monitor land cover and land change in the context of remote sensing. The full capabilities of a CNN were examined, including testing whether the platform could be used for land cover and the evaluation of land change over time. An unmanned aerial vehicle (UAV) was used to collect the aerial data of the study area. The CNN was encoded and operated in RStudio, while digitised data from the input imagery were used by the programme as training and validation data. The object in this respect was to learn about the relevant features of the landscape, and thereafter to classify the Opuntia invasive plant species. Accuracy assessments were carried out on the results to test the efficacy of the aerial imagery in terms of its accuracy and reliability. The classification achieved an overall accuracy of 93%, while the kappa coefficient score was 0.86. CNN was also able to predict the land coverage area of Opuntia to be within four percent (4%) of the ground truthing data. A change in land cover over time was detected by the programme after the manual clearing of the plant had been undertaken. This research has determined that the use of a CNN in remote sensing is a very powerful tool for supervised image classifications. It can be used for monitoring land cover in that it is able to accurately estimate the spatial distribution of plant species and to monitor the growth or decline in the species over time. As such, it is an efficient methodology and its use in remote sensing could be extended.

Construction of an Orthophoto-Draped 3D Model and Classification of Intertidal Habitats Using UAV Imagery in the Galapagos Archipelago

Worldwide, an increasing number of marine islands suffer from various pressures on the environment, driven by climate change and increasing land demands. The Galapagos Archipelago is one of the most iconic group of islands, yet population growth and tourism have resulted in a rising need for efficient environmental monitoring of its fragile ecosystems, such as the intertidal zone which harbors diverse and unique fauna. The purpose of this study was to investigate the image classification opportunities for these intertidal habitats using Uncrewed Aerial Vehicle (UAV) imagery. The data for this research were collected in Puerto Ayora on Santa Cruz in August 2017, the most urbanized island of the Galapagos. An orthophoto, a digital elevation model (DEM), and an orthophoto-draped 3D model of the intertidal zone were obtained using image registration software. Based on the orthophoto, an initial classification of the intertidal zone was performed using the spectral angle mapper algorithm. A habitat map with four classes (water, sand, rock, and vegetation) was created with an overall classification accuracy of 77%, indicating the suitability of UAV high resolution aerial imagery for the classification of intertidal habitats. The developed method could be applied to map and monitor other coastal regions and islands systems.

Decoding Nature's Patterns: An Innovative Approach to Tree Detection Using Deep Learning and High-Resolution Aerial Imagery

This study investigates the application of deep learning algorithms and high-resolution aerial imagery for individual tree detection in urban areas, using a neighborhood in Mersin, Turkey, as a case study. Employing the DeepForest Python package, we utilize high-resolution (7cm) aerial imagery to detect and map the city's tree population accurately. The results showcase an impressive accuracy rate of 80.87%, demonstrating the potential of deep learning in urban forestry applications and contributing to effective urban planning. The information generated from this study is crucial for conserving urban green spaces, enhancing resilience to climate change, and supporting urban biodiversity. While this research is focused on Mersin, the methods employed are globally adaptable, laying a foundation for further refinement and potential identification of different tree species in future work. This investigation highlights the transformative role of advanced technology in fostering sustainable urban environments.

Urbanisation of a growing tropical mega-city during the 21st century — Landscape transformation and vegetation dynamics

Fine scale spatial and temporal patterns in land-cover dynamics arising from rapid urbanisation of tropical regions are poorly understood. We quantify changes in landcover across the Bangkok region using high-resolution aerial imagery from ∼ 2004 to ∼ 2018 and address three questions: i) does urbanisation generate temporal shifts in the form of vegetation cover-urbanisation intensity relationships?, ii) do urban expansion and densification generate different vegetation dynamics?, iii) do net changes in vegetation cover and loss vary with urbanisation intensity? The form of vegetation cover-urbanisation intensity relationships exhibited negligible temporal variation, supporting the use of space-for-time substitution approaches for predicting future landcover dynamics. During our study period impervious surface cover increased by ∼ 474 km2and there were net losses of grassland (∼583 km2) and rice-fields (∼94 km2), and a net gain in treecover (∼137 km2). These changes have substantial implications for urban heat islands, flood risk and biodiversity. Urban expansion contributed more to vegetation dynamics than densification, partly because expansion impacted more land. Densification minimised loss of green-space, grasslands, and agriculture (rice-fields), but generated substantial local tree cover loss, which is critical to retain in highly urbanised areas. This contrasts with increasing tree cover elsewhere, including areas experiencing urban expansion. Trade-offs thus arise between impacts on different vegetation types when meeting the demand for tropical urban development through densification or expansion. Densification benefits most vegetation types but must be accompanied with tree protection and planting schemes to balance these trade-offs and minimise detrimental impacts of densification on people, ecosystem services and biodiversity.

Classification of objects within agriculture landscape from high resolution aerial imagery using maximum likelihood discriminant rules

In this study, the performance of object classification based on four discriminant functions, namely linear, quadratic, diagonal linear and diagonal quadratic is investigated and compared. High-resolution aerial imagery captured from a UAV-based remote sensing platform is used for this purpose. Initially, K-means clustering of 9 clusters is used to assist in the selection of training pixels for the subsequent supervised classification implementation. The classification is experimented with using a training set size of 10 through 100 pixels for each of the discriminant functions. The outcome of the classification shows that training set size 40 and 10 are to be the optimal training set sizes for linear and quadratic discriminant function, and diagonal linear and quadratic discriminant function respectively. Overall, the linear discriminant function is found to have the highest overall accuracy of 91% followed by diagonal linear, quadratic, and diagonal quadratic discriminant function with overall accuracies of 82%, 79%, and 73% respectively.