Search

This study estimates building attributes—floor area and age—using image‐based machine learning. Building age and floor area are key inputs to the studies of urban metabolism, material stocks and flows, and embodied greenhouse gases (GHGs) in the built environment. However, these data are challenging to generate and maintain using traditional survey methods, their availability is uneven and often, even when available, very uncertain. Improving our understanding and future management of built environment resource flows and associated environmental impacts requires more complete access to building age and floor area data. The study formulates area prediction as a regression problem and age prediction as a classification problem over six historical periods, achieving a mean absolute percentage error of 19.42% for area prediction and an accuracy of 70.27% for age prediction in Toronto. These results are obtained using an EfficientNetV2 module for feature extraction from Google Street View images, followed by fully connected layers for estimating the two building attributes. The performance of the Toronto‐trained model in five other Canadian cities is also reported, highlighting the model's varying effectiveness in different urban contexts and the benefit of local training. Our findings demonstrate the feasibility of using machine learning for building attribute estimation from street‐view images, offering a basis for future automated large‐scale material flow and stock analysis. [ABSTRACT FROM AUTHOR]

Issue Addressed Methods Results Conclusions So What? Food environments surrounding schools have a strong influence on the adolescent's food choices. Moreover, the prevalence of diet‐related chronic diseases is higher in regional than metropolitan areas in Australia. Understanding school food environments in these different settings is crucial for informing future strategies to improve adolescent health.Google Street View was used to identify food outlets within 1.6 km around all secondary schools in Wagga Wagga and Blacktown in New South Wales which were selected as regional and metropolitan case study areas. Based on food outlet type, healthfulness categories were assigned, and Chi‐squared tests were performed. The Health Advisory Panel for Youth at the University of Sydney (HAPYUS) were engaged to obtain their perspectives on findings.Unhealthful food outlets were consistently most prevalent around schools in Wagga Wagga and Blacktown over 17 years. In 2023, these were predominantly restaurants (19.4% vs. 21.1%), cafés (16.8% vs. 11.1%), fast‐food franchise outlets (15.1% vs. 17.4%) and independent takeaway stores (14.1% vs. 9.6%). No significant difference in healthfulness between regional and metropolitan areas was found. Youth advisors recognised price and social reasons as major contributors to food choices.Google Street View was used as a novel resource to examine school food environments in regional and metropolitan areas which have remained consistently unhealthful for nearly two decades.Unhealthful school food environments may encourage poor diets and exacerbate rates of adolescent overweight and obesity. Critical government action is needed to improve school food environments. [ABSTRACT FROM AUTHOR]

This research investigates the efficacy of OpenStreetMap (OSM) and Geographic Information Systems (GIS) for rapidly identifying and cataloging masonry bridges in specific regions integrating modern technology with historical data to enhance the mapping and preservation of these critical structures. The methodology developed is based on the use of OSM, Google Street View (GSV), and the Overpass API. It includes data extraction, processing, and validation through intrinsic and extrinsic quality measures, highlighting the robustness of masonry bridges, particularly in historical areas where traditional construction techniques prevail. The experimentation was carried out on an area of 1.225 km² in the province of Pescara (Italy). The results showed an accuracy rate of 74.35% in identifying bridge types from street-level images, despite the crowd-sourced nature of the data demonstrating the potential of combining open-access geospatial data with modern verification tools to support infrastructure mapping and preservation efforts, providing a comprehensive and cost-effective approach, especially in regions with deficient geospatial data. [ABSTRACT FROM AUTHOR]

• South Africa has a long history of introduction of trees from other parts of the world, starting with introductions in the mid-17th century. • Planted trees (both native and non-native) now dominate treescapes in many part of the country, especially in urban ecosystems. • Although planted trees are a key component of South Africa's green infrastructure and provide diverse ecosystem services, no up-date list of planted tree species exists. • This study used the citizen science platform iNaturalist to compile a spatially-explicit "living inventory" of planted tree species for South Africa. • The planted tree flora of South Africa comprises over 35,000 records of 805 taxa, 79 % of which are non-native. Over a third these taxa belong to three families: Fabaceae, Myrtaceae and Arecaceae. • Examples are provided of uses of the inventory for managing diverse aspects of South Africa's treescapes, including for monitoring the emergence of new invasions. Trees have a multifaceted influence on ecosystems globally. Treescapes have been manipulated by humans over millennia for ecological, economic, and cultural reasons that have changed over time and continue to change. In South Africa tree planting over three centuries has radically affected the composition of treescapes, contributing important ecosystem services, but also disservices. Rapid global change calls for diverse interventions to create more resilient ecosystems. Many nature-based solutions involve manipulating tree cover in rural and urban landscapes. There is a need for a spatially-explicit database of planted trees in South Africa to serve as the foundation for policy and management decisions. We used the community science platform iNaturalist to create a comprehensive database of planted trees in South Africa. Records were carefully checked to verify the accuracy of taxon identifications, locality data, and categorization as planted rather than wild-growing trees. The cleaned database contained 35,303 records of 805 planted tree taxa; over 90 % of records were identified to species level. Almost a third of taxa (32.2 %) belong to three families: Fabaceae (97 taxa), Myrtaceae (80 taxa) and Arecaceae (78 taxa). Rarefaction and extrapolation curves suggest fairly comprehensive sampling, but several regions are under-sampled. Non-native taxa dominate, with Melaleuca viminalis having the most records. The Western Cape has the highest number of records and taxa, particularly in urban areas. Both native and non-native trees provide key ecosystem services, but non-natives dominate. The database provides the first spatially-explicit open-access resource for guiding decisions on tree planting and the management of planted trees in South Africa. It offers a snapshot of tree planting trends, predominantly from recent years, leading to some underrepresentation of historically planted species. The database has numerous potential uses, including guiding management of trees pests and diseases, urban greening initiatives, monitoring for new invasions, and planning nature-based solutions. [ABSTRACT FROM AUTHOR]

An increased reliance on payment system technologies globally has led to the proliferation of card skimming as an unintended consequence. Moreover, skimming prevention has also proved challenging, particularly at lesser supervised public places like gas stations. The current study aims to determine if there are identifiable spatial patterns or environmental characteristics that predict skimming at gas stations. Density analyses and a remote systematic social observation (SSO) of Google Street View imagery were used to assess the environmental contexts associated with high skimming gas stations compared to controls. Results indicate place management features related to increased visibility and lines of sight significantly predict skimming risk. Policy and practice implications are discussed, as well as relevant applications to environmental criminology approaches explaining crime occurrence at the micro-level. [ABSTRACT FROM AUTHOR]

New data sources and AI methods for extracting information are increasingly abundant and relevant to decision-making across societal applications. A notable example is street view imagery, available in over 100 countries, and purported to inform built environment interventions (e.g., adding sidewalks) for community health outcomes. However, biases can arise when decision-making does not account for data robustness or relies on spurious correlations. To investigate this risk, we analyzed 2.02 million Google Street View (GSV) images alongside health, demographic, and socioeconomic data from New York City. Findings demonstrate robustness challenges; built environment characteristics inferred from GSV labels at the intracity level often do not align with ground truth. Moreover, as average individual-level behavior of physical inactivity significantly mediates the impact of built environment features by census tract, intervention on features measured by GSV would be misestimated without proper model specification and consideration of this mediation mechanism. Using a causal framework accounting for these mediators, we determined that intervening by improving 10% of samples in the two lowest tertiles of physical inactivity would lead to a 4.17 (95% CI 3.84-4.55) or 17.2 (95% CI 14.4-21.3) times greater decrease in the prevalence of obesity or diabetes, respectively, compared to the same proportional intervention on the number of crosswalks by census tract. This study highlights critical issues of robustness and model specification in using emergent data sources, showing the data may not measure what is intended, and ignoring mediators can result in biased intervention effect estimates. [ABSTRACT FROM AUTHOR]

This study explored the association between greening and social capital in neighborhoods, considering demographics and zoning by urban planning. The target area encompassed the urban areas of Kyoto City, Japan, which has a long tradition of landscape policy and neighborhood associations. Greening was evaluated using two automated methods: 1) horizontal green coverage of the land was calculated via the Normalized Difference Vegetation Index in satellite images, and 2) green visibility in streetscape from a human perspective was estimated by combining Google Street View images and a machine learning model. Public government data were used for demographics and zoning, and social capital was evaluated using survey data from the local government. After performing the elastic net models, variables that had explanatory power for each greening index were selected. Similar reasonable associations were found for each of the indices with the zoning categories. However, for both zoning and demographics, different variables were selected. Importantly, the social capital variable was selected only for the green visibility in streetscape, showing a negative correlation between them, unlike in previous studies. These results suggest that the association between urban greening and social relationships can change depending on the context of the target regions and measurements of greening.

Abstract Introduction Wearing a helmet reduces the risk of head injuries substantially in the event of a motorcycle crash. Countries around the world are committed to promoting helmet use, but the progress has been slow and uneven. There is an urgent need for large-scale data collection for situation assessment and intervention evaluation. Methods This study proposes a scalable, low-cost algorithm to estimate helmet-wearing rates. Applying the state-of-the-art deep learning technique for object detection to images acquired from Google Street View, the algorithm has the potential to provide accurate estimates at the global level. Results Trained on a sample of 3995 images, the algorithm achieved high accuracy. The out-of-sample prediction results for all three object classes (helmets, drivers, and passengers) reveal a precision of 0.927, a recall value of 0.922, and a mean average precision at 50 (mAP50) of 0.956. Discussion The remarkable model performance suggests the algorithm’s capacity to generate accurate estimates of helmet-wearing rates from an image source with global coverage. The significant enhancement in the availability of helmet usage data resulting from this approach could bolster progress tracking and facilitate evidence-based policymaking for helmet wearing globally.

Introduction: Wearing a helmet reduces the risk of head injuries substantially in the event of a motorcycle crash. Countries around the world are committed to promoting helmet use, but the progress has been slow and uneven. There is an urgent need for large-scale data collection for situation assessment and intervention evaluation. Methods: This study proposes a scalable, low-cost algorithm to estimate helmet-wearing rates. Applying the state-of-the-art deep learning technique for object detection to images acquired from Google Street View, the algorithm has the potential to provide accurate estimates at the global level. Results: Trained on a sample of 3995 images, the algorithm achieved high accuracy. The out-of-sample prediction results for all three object classes (helmets, drivers, and passengers) reveal a precision of 0.927, a recall value of 0.922, and a mean average precision at 50 (mAP50) of 0.956. Discussion: The remarkable model performance suggests the algorithm's capacity to generate accurate estimates of helmet-wearing rates from an image source with global coverage. The significant enhancement in the availability of helmet usage data resulting from this approach could bolster progress tracking and facilitate evidence-based policymaking for helmet wearing globally. [ABSTRACT FROM AUTHOR]

Street-level image platforms (SLIPs) employ indiscriminate forms of data collection that include potentially privacy invasive images. Both the scale and the indiscriminate nature of data collection means that significant privacy management requirements are needed. Legal risk management is currently operated through obfuscation techniques involving certain image objects. Current SLIP object obfuscation solutions are an indiscriminate and a blunt solution to a similarly indiscriminate data collection concern. A new contextual approach to obfuscation is required that goes beyond object obfuscation. Contextually-dependent identification would seek to identify the contexts, including captured objects, which can give rise to privacy concerns. It is technically more challenging for automated solutions as it requires an assessment of the contextual situation to understand privacy risk. Context-sensitive privacy detection, combined with context-sensitive privacy-by-design processes, potentially offer a risk management solution that better situates and addresses the concerns arising from SLIP data collections. [ABSTRACT FROM AUTHOR]

Flood events can cause extensive damage to physical infrastructure, pose risks to human life, and necessitate the reoccupation and rehabilitation of affected areas. A key parameter for flood vulnerability assessment is the first floor height (FFH), which also plays an important role in setting insurance premiums. Traditional methods for FFH estimation rely on ground surveys and site inspections, yet these approaches are both time‐consuming and labor‐intensive. In this study, we propose an alternative approach based on measurements derived from Google Street View (GSV) images and Deep Learning (DL). We employ the YOLOv5s algorithm, which belongs to a family of compound‐scaled object detection models trained on the COCO dataset, for the detection of crucial building elements such as the Front Door (FD), stairs, and overall building extent. Additionally, we utilized the YOLOv5s algorithm to identify basement windows and assess the existence of basements. To validate our methodology, we conducted tests in both the Greater Toronto Area (GTA) and the state of Virginia in the United States. The results demonstrate an achievement of RMSE and Bias values of 81 cm and −50 cm for GTA, and 95 cm and −20 cm for the Virginia region, respectively. [ABSTRACT FROM AUTHOR]

Background and Aims Built environment plays an important role in the development of cardiovascular disease. Tools to evaluate the built environment using machine vision and informatic approaches have been limited. This study aimed to investigate the association between machine vision–based built environment and prevalence of cardiometabolic disease in US cities. Methods This cross-sectional study used features extracted from Google Street View (GSV) images to measure the built environment and link them with prevalence of coronary heart disease (CHD). Convolutional neural networks, linear mixed-effects models, and activation maps were utilized to predict health outcomes and identify feature associations with CHD at the census tract level. The study obtained 0.53 million GSV images covering 789 census tracts in seven US cities (Cleveland, OH; Fremont, CA; Kansas City, MO; Detroit, MI; Bellevue, WA; Brownsville, TX; and Denver, CO). Results Built environment features extracted from GSV using deep learning predicted 63% of the census tract variation in CHD prevalence. The addition of GSV features improved a model that only included census tract-level age, sex, race, income, and education or composite indices of social determinant of health. Activation maps from the features revealed a set of neighbourhood features represented by buildings and roads associated with CHD prevalence. Conclusions In this cross-sectional study, the prevalence of CHD was associated with built environment factors derived from GSV through deep learning analysis, independent of census tract demographics. Machine vision–enabled assessment of the built environment could potentially offer a more precise approach to identify at-risk neighbourhoods, thereby providing an efficient avenue to address and reduce cardiovascular health disparities in urban environments. [ABSTRACT FROM AUTHOR]

The recent advent of virtual photography and artificial intelligence (specifically AI photography and diffusion models), presents a major challenge to both photographers and photography theory. All our common-sense assumptions about the ontology of the image now seem to be falling apart. Are these interlopers really photography? Can we hear the final death knell for the erstwhile medium in this era of ever-increasing media convergence and virtualisation? This practice-led investigation of these new media forms draws on the author's recent creative work in virtual photographies, and similar interventions by other photographers and new media artists, seeking to augment and expand their practice using these new tools, whilst querying what photography really means today. From in-game photography to virtual exploration using Google Street View, to AI photography using the latest denoising diffusion models (such as Midjourney), there are surprising commonalities to explore between them, linking these new practices of image-making firmly back to traditions of lens-based photography. Rather than seeking a detailed map of this difficult new terrain, or a definitive ontology of emerging virtual photographies, the author reframes the discourse around practice, examining both photography as an evolving set of practices and also notions of media hauntology – specifically the spectral ways in which new media technologies are always haunted by prior practices and modes of communication. The importance of the frame to both traditional photography and new practices of virtual photography, is suggested as a vital and persistent dimension in photographic authorship. With that authorship increasingly contested by new generative methods, easy appropriation and AI image-making, the act of framing may become the best litmus test we have, for whether or not a photograph should be considered 'real' or valued. [ABSTRACT FROM AUTHOR]

Most image positioning systems use the static angle of view of the query image and compare it with images containing geographical information in the database to deduce its location. However, in some cases, the location information of the target object in the image is equally important, except for the position of the camera at that time. For example, when a disaster is reported, if an image can be utilized to estimate its precise location, it can replace the inaccurate description of the target point in the report, which will greatly improve the speed of rescue. We propose a three-dimensional positioning system for the point of interest in an image. In order to improve the positioning accuracy, this system integrates the Global Positioning System location and Google Maps data to obtain information regarding the point of interest and infer the three-dimensional coordinate of the target within the user's sight by triangulation. We have conducted experiments using several known points to achieve a horizontal precision of 1.013 m with a root mean square error of 1.029 m and a vertical accuracy of 0.962 m with a root mean square error of 1.329 m. [ABSTRACT FROM AUTHOR]

Objectives: Despite theoretical interest in how dimensions of the built environment can help explain the location of crime in micro−geographic units, measuring this is difficult. Methods: This study adopts a strategy that first scrapes images from Google Street View every 20 meters in every street segment in the city of Santa Ana, CA, and then uses machine learning to detect features of the environment. We capture eleven different features across four main dimensions, and demonstrate that their relative presence across street segments considerably increases the explanatory power of models of five different Part 1 crimes. Results: The presence of more persons in the environment is associated with higher levels of crime. The auto−oriented measures—vehicles and pavement—were positively associated with crime rates. For the defensible space measures, the presence of walls has a slowing negative relationship with most crime types, whereas fences did not. And for our two greenspace measures, although terrain was positively associated with crime rates, vegetation exhibited an inverted−U relationship with two crime types. Conclusions: The results demonstrate the efficacy of this approach for measuring the built environment.

Urban visual localization is the process of determining the pose (position and attitude) of the imaging sensor (or platform) with the help of existing geo-referenced data. This task is critical and challenging for many applications, such as autonomous navigation, virtual and augmented reality, and robotics, due to the dynamic and complex nature of urban environments that may obstruct Global Navigation Satellite Systems (GNSS) signals. This paper proposes a block-wise matching strategy for urban visual localization by using geo-referenced Google Street View (GSV) panoramas as the database. To determine the pose of the monocular query images collected from a moving vehicle, neighboring GSVs should be found to establish the correspondence through image-wise and block-wise matching. First, each query image is semantically segmented and a template containing all permanent objects is generated. The template is then utilized in conjunction with a template matching approach to identify the corresponding patch from each GSV image within the database. Through the conversion of the query template and corresponding GSV patch into feature vectors, their image-wise similarity is computed pairwise. To ensure reliable matching, the query images are temporally grouped into query blocks, while the GSV images are spatially organized into GSV blocks. By using the previously computed image-wise similarities, we calculate a block-wise similarity for each query block with respect to every GSV block. A query block and its corresponding GSV blocks of top-ranked similarities are then input into a photogrammetric triangulation or structure from motion process to determine the pose of every image in the query block. A total of three datasets, consisting of two public ones and one newly collected on the Purdue campus, are utilized to demonstrate the performance of the proposed method. It is shown it can achieve a meter-level positioning accuracy and is robust to changes in acquisition conditions, such as image resolution, scene complexity, and the time of day. [ABSTRACT FROM AUTHOR]

Biological invasions threaten global biodiversity and can have severe economic and social impacts. The complexity of this problem challenges effective management of invasive alien species as the contribution of many factors involved in the invasion processes across different spatial scales is not well understood.Here, we identify the most important determinants associated with the occurrence of two invasive alien plants, the North American goldenrods (Solidago canadensis and S. gigantea), commonly found in agricultural landscapes of Europe. We used Google Street View images to perform a remote, large‐scale inventory of goldenrods along 1347 roadside transects across Poland. Using open access geospatial data and machine learning techniques, we investigated the relative role of nearly 50 variables potentially affecting the distribution of studied species at five spatial scales (from within 0.25 to 5 km of the studied locations).We found that the occurrence of goldenrods along roadsides was simultaneously associated with multiple drivers among which those related to human impacts, climate, soil properties and landscape structure were the most important, while local characteristics, such as road parameters or the presence of other alien plants were less influential. However, the relative contribution of different variables in predicting goldenrod distribution changed across spatial scales.Synthesis: Mechanisms underlying plant invasions are highly complex and a number of factors can jointly influence the outcomes of this process. However, since different invasion drivers operate at different spatial scales, some important associations may be overlooked when focusing on a single spatial context. Although associations were consistent in direction (positive or negative) across scales, their relative influence on goldenrod occurrence often changed. Socio‐economic factors were largely important at local scales, while the effect of landscape factors broadly increased with increasing spatial scale. We highlight that using multi‐scale approaches involving a wide range of variables may enable setting priorities for the management of invasive alien plants. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Teaching Practice & Pedagogical Innovation is the property of National Taipei University of Education and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Global localization. Google Earth and the experience of place attachment in GeoGuessr, Houellebecq and Darrieussecq The purpose of this article is to analyze literary texts and other cultural practices that relate to or use content from Google Earth and, more generally, to outline what I call a contemporary glocal experience of place. The article discusses the browser game GeoGuessr and the interactive music video “The Wilderness Downtown” by Arcade Fire, along with the novel La Mer à l’envers by Marie Darrieussecq, and different works, including photographs and poems, by Michel Houellebecq. In all these instances, the transmissions, slippages, and tensions between a situated and embodied experience of place and global technological systems play a prominent role. The article shows how GeoGuessr and “The Wilderness Downtown” explore this tension from opposite points of departure: while the game fosters a sense of place within locales that are essentially unknown to the player, the music video integrates images from places that are related to the viewer’s personal memories into an otherwise impersonal animated video. In turn, Darrieussecq’s novel combines and narrativizes these opposing perspectives. Here, Calais represents a globalized space that can only be understood through glocalizing technologies, such as geoposititioning and interactive mapping. But Clève, the hometown of the protagonist, becomes unmoored through the same interaction with these technologies. Finally, the article highlights the “Houellebecq perspective” in novels, poems, and photographs. In these recurring scenes, the world is seen from above (through satellite imagery, on maps, or through airplanes windows). They revolve around the tension between the aerial view and the “horizon structure”, which permits the subject to place itself in the world and in relation to others.

Abstract Flood events can cause extensive damage to physical infrastructure, pose risks to human life, and necessitate the reoccupation and rehabilitation of affected areas. A key parameter for flood vulnerability assessment is the first floor height (FFH), which also plays an important role in setting insurance premiums. Traditional methods for FFH estimation rely on ground surveys and site inspections, yet these approaches are both time‐consuming and labor‐intensive. In this study, we propose an alternative approach based on measurements derived from Google Street View (GSV) images and Deep Learning (DL). We employ the YOLOv5s algorithm, which belongs to a family of compound‐scaled object detection models trained on the COCO dataset, for the detection of crucial building elements such as the Front Door (FD), stairs, and overall building extent. Additionally, we utilized the YOLOv5s algorithm to identify basement windows and assess the existence of basements. To validate our methodology, we conducted tests in both the Greater Toronto Area (GTA) and the state of Virginia in the United States. The results demonstrate an achievement of RMSE and Bias values of 81 cm and −50 cm for GTA, and 95 cm and −20 cm for the Virginia region, respectively.

Criminological theories have posited that the built environment impacts where crime occurs; however, measuring the built environment is difficult. Furthermore, it is uncertain whether the built environment differentially impacts crime in high-disadvantage neighborhoods. This study extracts features of the built environment from Google Street View images with a machine learning semantic segmentation strategy to create measures of fences, walls, buildings, and greenspace for over 66,000 street segments in Los Angeles. Results indicate that the presence of more buildings on a segment was associated with higher crime rates and had a particularly strong positive relationship with robbery and motor vehicle theft in low-disadvantage neighborhoods. Notably, fences and walls exhibited different relationships with crime. Walls, which do not allow visibility, were strongly negatively related to crime, particularly for robbery and burglary in high-disadvantage neighborhoods. Fences, which allow visibility, were associated with fewer robberies and larcenies, but more burglaries and aggravated assaults. Fences only exhibited a negative relationship with violent crime when they were located in low-disadvantage neighborhoods. The results highlight the importance of accounting for the built environment and the surrounding level of disadvantage when exploring the micro-location of crime.

Characteristics of the neighborhood built environment influence health and health behavior. Google Street View (GSV) images may facilitate measures of the neighborhood environment that are meaningful, practical, and adaptable to any geographic boundary. We used GSV images and computer vision to characterize neighborhood environments (green streets, visible utility wires, and dilapidated buildings) and examined cross-sectional associations with chronic health outcomes among patients from the University of California, San Francisco Health system with outpatient visits from 2015 to 2017. Logistic regression models were adjusted for patient age, sex, marital status, race/ethnicity, insurance status, English as preferred language, assignment of a primary care provider, and neighborhood socioeconomic status of the census tract in which the patient resided. Among 214,163 patients residing in California, those living in communities in the highest tertile of green streets had 16-29% lower prevalence of coronary artery disease, hypertension, and diabetes compared to those living in communities in the lowest tertile. Conversely, a higher presence of visible utility wires overhead was associated with 10-26% more coronary artery disease and hypertension, and a higher presence of dilapidated buildings was associated with 12-20% greater prevalence of coronary artery disease, hypertension, and diabetes. GSV images and computer vision models can be used to understand contextual factors influencing patient health outcomes and inform structural and place-based interventions to promote population health.

Urban tree inventories have mostly focused on the information of individual trees becausethese allows city authorities to efficiently plan urban forestation. However, single-tree urban tree inventories are expensive for municipalities, so the inventories lack detail and are often out of date. In this work, we aim to integrate the possibility of using online applications for automatic species identification with worldwide coverage Pl@ntNet and Plant.Id on Google Street View (GSV) images in order to perform cost-effective urban tree inventories at the single-tree level and evaluate the performance of the two applications through comparison with a locally trained neural network using an appropriate set of metrics. Our work showed that the Plant.Id application gave the best performance by correctly identifying plants in the city of Prato with a median accuracy of 0.73 and better performance for the most common plants: Pinus pinea 0.87, Tilia aeuropea 0.87, Platanus hybrida 0.89. The proposed method also has a limitation. Trees within parks, walking paths and private green areas cannot be photographed and identified because Google cars cannot access them. The solution to this limitation is to combine GSV images with spherical photos taken via light unmanned aircraft. [ABSTRACT FROM AUTHOR]

A widely used method for estimating fine scale long-term spatial variation in air pollution, especially for epidemiology studies, is land use regression (LUR) modeling using fixed off-road monitors. More recently, LUR models have been developed using data from mobile monitors that repeatedly measure road pollutants and mixed-effects modeling. Here, nitrogen dioxide (NO2) and black carbon (BC) predictions from two independent models were compared across streets (defined as 30–60 m road segments) (N = 30,312) and residences (N = 76,752) in Copenhagen, Denmark. The first model was Google Street View (GSV)-based mixed-effects LUR models (Google-MM) that predicted 2019 mean NO2 and BC levels, and the second was European-wide (EUW) LUR models that predicted annual mean 2010 levels at 100 m spatial resolution. Across street segments, the Spearman correlation coefficient between the 2019 NO2 from Google-MM-LUR and 2010 NO2 from EUW-LUR was 0.66, while at residences, this was 0.60. For BC, these were 0.51 across street segments and 0.40 at the residential level. The ratio of percentile 97.5 to 2.5 for NO2 across the study area streets using Google-MM NO2 was 4.5, while using EUW-LUR, this was 2.1. These NO2 ratios at residences were 3.1 using Google-MM LUR, and 1.7 using EUW-LUR. Such ratios for BC across street segments were 3.4 using Google-MM LUR and 2.3 using EUW-LUR, while at the residential level, they were 2.4 and 1.9, respectively. In conclusion, Google-MM-LUR NO2 for 2019 was moderately correlated with EUW-LUR NO2 developed in 2010 across Copenhagen street segments and residences. For BC, while Google-MM-LUR was moderately correlated with EUW-LUR across Copenhagen streets, the correlation was lower at the residential level. Overall, Google-MM-LUR revealed larger spatial contrasts than EUW-LUR. [ABSTRACT FROM AUTHOR]

While Google Street View (GSV) has been increasingly available for large-scale examinations of urban landscapes, little is known about how to use this promising data source more cautiously and effectively. Using data for Santa Ana, California, as an example, this study provides an empirical assessment of the sensitivity of GSV-based streetscape measures and their variation patterns. The results show that the measurement outcomes can vary substantially with changes in GSV acquisition parameter settings, specifically spacing and direction. The sensitivity is found to be particularly high for some measurement targets, including humans, objects, and sidewalks. Some of these elements, such as buildings and sidewalks, also show highly correlated patterns of variation indicating their covariance in the mosaic of urban space.

While Google Street View (GSV) has been increasingly available for large-scale examinations of urban landscapes, little is known about how to use this promising data source more cautiously and effectively. Using data for Santa Ana, California, as an example, this study provides an empirical assessment of the sensitivity of GSV-based streetscape measures and their variation patterns. The results show that the measurement outcomes can vary substantially with changes in GSV acquisition parameter settings, specifically spacing and direction. The sensitivity is found to be particularly high for some measurement targets, including humans, objects, and sidewalks. Some of these elements, such as buildings and sidewalks, also show highly correlated patterns of variation indicating their covariance in the mosaic of urban space.

Objectives: Despite theoretical interest in how dimensions of the built environment can help explain the location of crime in micro−geographic units, measuring this is difficult. Methods: This study adopts a strategy that first scrapes images from Google Street View every 20 meters in every street segment in the city of Santa Ana, CA, and then uses machine learning to detect features of the environment. We capture eleven different features across four main dimensions, and demonstrate that their relative presence across street segments considerably increases the explanatory power of models of five different Part 1 crimes. Results: The presence of more persons in the environment is associated with higher levels of crime. The auto−oriented measures—vehicles and pavement—were positively associated with crime rates. For the defensible space measures, the presence of walls has a slowing negative relationship with most crime types, whereas fences did not. And for our two greenspace measures, although terrain was positively associated with crime rates, vegetation exhibited an inverted−U relationship with two crime types. Conclusions: The results demonstrate the efficacy of this approach for measuring the built environment.

ABSTRACTUrban tree inventories have mostly focused on the information of individual trees becausethese allows city authorities to efficiently plan urban forestation . However, single-tree urban tree inventories are expensive for municipalities, so the inventories lack detail and are often out of date. In this work, we aim to integrate the possibility of using online applications for automatic species identification with worldwide coverage Pl@ntNet and Plant.Id on Google Street View (GSV) images in order to perform cost-effective urban tree inventories at the single-tree level and evaluate the performance of the two applications through comparison with a locally trained neural network using an appropriate set of metrics. Our work showed that the Plant.Id application gave the best performance by correctly identifying plants in the city of Prato with a median accuracy of 0.73 and better performance for the most common plants: Pinus pinea 0.87, Tilia aeuropea 0.87, Platanus hybrida 0.89. The proposed method also has a limitation. Trees within parks, walking paths and private green areas cannot be photographed and identified because Google cars cannot access them. The solution to this limitation is to combine GSV images with spherical photos taken via light unmanned aircraft.

This paper aims to address the challenges associated with evaluating the impact of neighborhood environments on health outcomes. Google street view (GSV) images provide a valuable tool for assessing neighborhood environments on a large scale. By annotating the GSV images with labels indicating the presence or absence of specific neighborhood features, we can develop classifiers capable of automatically analyzing and evaluating the environment. However, the process of labeling GSV images to analyze and evaluate the environment is a time-consuming and labor-intensive task. To overcome these challenges, we propose using a multi-task classifier to enhance the training of classifiers with limited supervised GSV data. Our multi-task classifier utilizes readily available, inexpensive online images collected from Flickr as a related classification task. The hypothesis is that a classifier trained on multiple related tasks is less likely to overfit to small amounts of training data and generalizes better to unseen data. We leverage the power of multiple related tasks to improve the classifier’s overall performance and generalization capability. Here we show that, with the proposed learning paradigm, predicted labels for GSV test images are more accurate. Across different environment indicators, the accuracy, $F_{1}$ score and balanced accuracy increase up to 6% in the multi-task learning framework compared to its single-task learning counterpart. The enhanced accuracy of the predicted labels obtained through the multi-task classifier contributes to a more reliable and precise regression analysis determining the correlation between predicted built environment indicators and health outcomes. The $R^{2}$ values calculated for different health outcomes improve by up to 4% using multi-task learning detected indicators.

Pedestrian interventions for healthier and more inclusive streetscapes can be powerful mechanisms to increase the safety and comfort of walking in African cities. This article proposes a multiscale walkability analysis approach to identify both suitable streets for pedestrian travel and problematic areas requiring small-scale improvements (e.g., pavement repairs, building maintenance, streetlights, and public seating). We applied a GIS-based framework to the central urban area of Cape Town, South Africa, which presents complex social and environmental challenges. For each street-and-crossing segment, a virtual pedestrian streetscape audit tool was used to collect micro- and mesoscale environmental indicators and assess the quality of public space. This composite street-level assessment tool was weighted with a space syntax analysis indicator (i.e., spatial integration) to detect the network’s most interconnected and high-priority pathways. The Jenks natural breaks classification algorithm was used to classify scores for each segment, which ultimately found that the highest-priority streets for redevelopment are clustered in Bo-Kaap, a relatively disadvantaged, multicultural, and hilly district on Cape Town’s west side. Policy recommendations are evaluated to increase the quality of the urban environment and the city’s overall attractiveness to pedestrians. The proposed methodology facilitates more effective place management and classifies the city’s needs in improvements, minimizing both time and budget costs.

Prostorsko urejanje območij za pešce, ki zagotavlja bolj zdravo in vključujočo ulično krajino, je lahko močan mehanizem za izboljšanje varnosti in udobnosti pešačenja v afriških mestih. Avtorja v članku predlagata pristop k analizi hodljivosti na več prostorskih ravneh, s katerim se lahko določijo ulice, primerne za pešce, in problematična območja, ki zahtevajo manjše izboljšave (npr. popravilo pločnikov, boljše vzdrževanje stavb ter ureditev ulične razsvetljave in javnih klopi). Raziskovalni okvir, ki temelji na uporabi GIS, sta uporabila za središče Cape Towna v Južni Afriki, ki se spopada s kompleksnimi družbenimi in okoljskimi izzivi. Za vsak segment ulice s prehodom za pešce sta z orodjem za virtualno presojo območij za pešce zbrala podatke o okoljskih kazalnikih na mikro- in mezoravni ter proučila kakovost javnega prostora. Rezultate, dobljene z navedenim sestavljenim orodjem za presojo ulic, sta pomnožila z normalizirano vrednostjo mere prostorske sintakse (tj. integracije), na podlagi česar sta določila poti v mreži, ki so med seboj najbolj povezane in najbolj potrebne prenove. Z Jenksovo metodo naravnih mej sta razvrstila rezultate za vsak segment, na podlagi česar sta ugotovila, da so ulice, ki so najbolj potrebne prenove, zgoščene v Bo-Kaapu, razmeroma slabo razvitem, večkulturnem in hribovitem predelu v zahodnem delu Cape Towna. Na koncu sta predstavila priporočila za izboljšanje kakovosti mestnega okolja in splošne privlačnosti mesta za pešce. Predlagana metodologija omogoča učinkovitejše upravljanje krajev in razvrščanje potreb mesta po izboljšavah, s čimer se zmanjšajo stroški in poraba časa.

Detecting asbestos-containing roofs has been of great interest in the past few years as the substance negatively affects human health and the environment. Different remote sensing data have been successfully used for this purpose. However, RGB and thermal data have yet to be investigated. This study aims to investigate the classification of asbestos-containing roofs using RGB and airborne thermal data and state-of-the-art machine learning (ML) classification techniques. With the rapid development of ML reflected in this study, we evaluate three classifiers: Random Forest (RF), Support Vector Machine (SVM), and eXtreme Gradient Boosting (XGBoost). We have used several image enhancement techniques to produce additional bands to improve the classification results. For feature selection, we used the Boruta technique; based on the results, we have constructed four different variations of the dataset. The results showed that the most important features for asbestos-containing roof detection were the investigated spectral indices in this study. From a ML point of view, SVM outperformed RF and XGBoost in the dataset using only the spectral indices, with a balanced accuracy of 0.93. Our results showed that RGB bands could produce as accurate results as the multispectral and hyperspectral data with the addition of spectral indices. [ABSTRACT FROM AUTHOR]

Urban visual localization is the process of determining the pose (position and attitude) of the imaging sensor (or platform) with the help of existing geo-referenced data. This task is critical and challenging for many applications, such as autonomous navigation, virtual and augmented reality, and robotics, due to the dynamic and complex nature of urban environments that may obstruct Global Navigation Satellite Systems (GNSS) signals. This paper proposes a block-wise matching strategy for urban visual localization by using geo-referenced Google Street View (GSV) panoramas as the database. To determine the pose of the monocular query images collected from a moving vehicle, neighboring GSVs should be found to establish the correspondence through image-wise and block-wise matching. First, each query image is semantically segmented and a template containing all permanent objects is generated. The template is then utilized in conjunction with a template matching approach to identify the corresponding patch from each GSV image within the database. Through the conversion of the query template and corresponding GSV patch into feature vectors, their image-wise similarity is computed pairwise. To ensure reliable matching, the query images are temporally grouped into query blocks, while the GSV images are spatially organized into GSV blocks. By using the previously computed image-wise similarities, we calculate a block-wise similarity for each query block with respect to every GSV block. A query block and its corresponding GSV blocks of top-ranked similarities are then input into a photogrammetric triangulation or structure from motion process to determine the pose of every image in the query block. A total of three datasets, consisting of two public ones and one newly collected on the Purdue campus, are utilized to demonstrate the performance of the proposed method. It is shown it can achieve a meter-level positioning accuracy and is robust to changes in acquisition conditions, such as image resolution, scene complexity, and the time of day.

As Google Street View visually depicts areas with disparate social characteristics, we use them to analyze the effects of environmentally locational factors on housing prices by constructing a convolutional neural network model. Instead of manual classification and judgment, the model decomposes views' pixels then assigns latent scores for street views. This score factor can improve the interpretability and the prediction accuracy of hedonic models and machine learning models. We empirically show this score is statistically significant and has stronger predictive power, suggesting that Google Street View provides visual cues regarding the dwelling's location and improve the regional and housing research. [ABSTRACT FROM AUTHOR]

Power and communication line corridors are usually mixed with urban trees, and this mixing can be the source of multiple issues like fires and communication failures. Nevertheless, urban trees are a valuable resource to the city as they dissipate heat island effects, reduce air pollution and increase general health perception. This work proposes a deep learning approach to detect trees entangled to power and communication lines using street-level imagery and perform quick quantitative and qualitative analyses based on the Grad-CAM++ method. Testing the method was performed using 1001 images from urban trees from the cities of São Paulo and Porto Alegre (both in Brazil). We found an overall accuracy of 74.6% (73.6% for São Paulo and 75.6% for Porto Alegre), suggesting that the methodology could be suitable in the future for city management to avoid risks of accidents due to contact between trees and electrical wiring. This text describes the method, a new data set of urban images, the experimental setup design and tests, and some possible future improvements. [ABSTRACT FROM AUTHOR]

Introduction: The goal of these studies was to investigate the reliability and validity of virtual systematic social observation (virtual SSO) using Google Street View in a Swedish neighborhood context. Methods: This was accomplished in two studies. Study 1 focused on interrater reliability and construct validity, comparing ratings conducted in-person to those done using Google Street View, across 24 study sites within four postal code areas. Study 2 focused on criterion validity of virtual SSO in terms of neighborhoods with low versus high income levels, including 133 study sites within 22 postal code areas in a large Swedish city. In both studies, assessment of the neighborhood context was conducted at each study site, using a protocol adapted to a Swedish context. Results: Scales for Physical Decay, Neighborhood Dangerousness, and Physical Disorder were found to be reliable, with adequate interrater reliability, high consistency across methods, and high internal consistency. In Study 2, significantly higher levels of observed Physical Decay, Neighborhood Dangerousness, and signs of garbage or litter were observed in postal codes areas (site data was aggregated to postal code level) with lower as compared to higher income levels. Discussion: We concluded that the scales within the virtual SSO with Google Street View protocol that were developed in this series of studies represents a reliable and valid measure of several key neighborhood contextual features. Implications for understanding the complex person-context interactions central to many theories of positive development among youth were discussed in relation to the study findings. [ABSTRACT FROM AUTHOR]

Bars have an established relationship to crime and are routinely operationalized as an important predictor of crime occurrence. However, despite this reputation, an interesting paradox exists in that most bars are not criminogenic. This study attempts to explain the variation in crime levels at bars by observing their environments over time. Environmental features of bars in Denver were recorded using year-over-year Google Street View imagery from 2014 to 2022. Analyses then examined the presence, predictivity, and patterning of the observed environmental features to better explain the relationship between bars, bar environments, and crime over time. The results indicate that a high number of environmental features are present at bars but only a few features significantly predict crime occurrence. The variation in crime levels at bars may be best explained through specific, situational environmental features and place management tactics. Though, identifying the temporal patterning of environmental features as static or dynamic over time is critical to understanding crime occurrence at bars. The results suggest that crime at bars may be attributable to unique combinations of environmental features and temporal considerations at individual bars. Explaining the variation in crime levels may require facility-by-facility nuance to better inform situational crime prevention efforts. • There is variation in crime levels across bars within a jurisdiction. • Environmental predictors of crime are frequently recorded at bars. • Only a select few environmental features predict crime at bars. • The temporal pattern of environmental predictors as static or dynamic over time helps explain the variation in crime at bars. • Situational crime prevention should priortize individual facility-level assessments of the environment and crime over time. [ABSTRACT FROM AUTHOR]

Spending time outdoors may bolster mental health via relaxation and physical activity. Yet, most studies use self-report and involve majority White samples from higher income areas. Findings may not hold true using device-derived measurement, among minoritized participants, or in low-income neighborhoods. Our objectives were to test whether: i) time spent outdoors in physical activity is associated with mental health; ii) the effect is modified by the positive or negative outdoor conditions (amount of greenness or vacant houses/lots encountered while outdoors); iii) there is a correlation between our measures of outdoor conditions and perceived neighborhood conditions; and iv) there is a different relationship between time spent outdoors in parks (as an indicator of intentional nature exposure) and mental health. The sample of 112 participants (87 % Black/African American persons) from 11 Detroit, MI neighborhoods wore a GPS and accelerometer for one week (May-Oct 2021) and completed a survey. We calculated exposure to greenness, vacant lots/houses while outdoors, and perceived neighborhood aesthetics. We fitted regression models for mental health measures separately and included time spent outdoors in PA, age, sex, and days with high precipitation. We then fitted models stratified by outdoor conditions. Last, we examined whether a similar relationship could be detected for time spent outdoors in parks. On average, ∼30 % of PA was conducted outdoors, and 56 % of outdoor time involved PA. We observed an unexpected positive association between outdoor PA and anxiety (coef: 0.33, 95 % CI: 0.14–0.51), depression (coef: 0.31, 95 % CI: 0.15–0.47), and perceived stress score (coef: 0.18, 95 % CI: 0.07–0.29). When stratified, we found consistent and/or larger associations for those exposed to high levels of vacant homes, high greenness (both aerial and street-level), low levels of vacant lots, and low perceived neighborhood aesthetics. Importantly and in contrast, we observed the opposite trend when examining time spent in parks (coef: −0.05, 95 % CI: −0.10--0.01). These results contrast many studies showing positive relationships between time outdoors and mental health and highlight the need for research with diverse populations and better measures of outdoor conditions and intentionality of greenspace contact. • We explored relationships between outdoor PA and mental health, and modification by outdoor conditions. • We found a positive association between outdoor PA and three measures of mental health. • We found consistent associations for those exposed to more vacant homes and greenery, fewer vacant lots, and low aesthetics. • In contrast, we observed the opposite trend when examining time spent in parks. • Research is needed in diverse populations, on better metrics of greenspace quality and on intentional greenspace contact. [ABSTRACT FROM AUTHOR]

A widely used method for estimating fine scale long-term spatial variation in air pollution, especially for epidemiology studies, is land use regression (LUR) modeling using fixed off-road monitors. More recently, LUR models have been developed using data from mobile monitors that repeatedly measure road pollutants and mixed-effects modeling. Here, nitrogen dioxide (NO2) and black carbon (BC) predictions from two independent models were compared across streets (defined as 30–60 m road segments) (N = 30,312) and residences (N = 76,752) in Copenhagen, Denmark. The first model was Google Street View (GSV)-based mixed-effects LUR models (Google-MM) that predicted 2019 mean NO2 and BC levels, and the second was European-wide (EUW) LUR models that predicted annual mean 2010 levels at 100 m spatial resolution. Across street segments, the Spearman correlation coefficient between the 2019 NO2 from Google-MM-LUR and 2010 NO2 from EUW-LUR was 0.66, while at residences, this was 0.60. For BC, these were 0.51 across street segments and 0.40 at the residential level. The ratio of percentile 97.5 to 2.5 for NO2 across the study area streets using Google-MM NO2 was 4.5, while using EUW-LUR, this was 2.1. These NO2 ratios at residences were 3.1 using Google-MM LUR, and 1.7 using EUW-LUR. Such ratios for BC across street segments were 3.4 using Google-MM LUR and 2.3 using EUW-LUR, while at the residential level, they were 2.4 and 1.9, respectively. In conclusion, Google-MM-LUR NO2 for 2019 was moderately correlated with EUW-LUR NO2 developed in 2010 across Copenhagen street segments and residences. For BC, while Google-MM-LUR was moderately correlated with EUW-LUR across Copenhagen streets, the correlation was lower at the residential level. Overall, Google-MM-LUR revealed larger spatial contrasts than EUW-LUR.

Pedestrian interventions for healthier and more inclusive streetscapes can be powerful mechanisms to increase the safety and comfort of walking in African cities. This article proposes a multiscale walkability analysis approach to identify both suitable streets for pedestrian travel and problematic areas requiring small-scale improvements (e.g., pavement repairs, building maintenance, streetlights, and public seating). We applied a GIS-based framework to the central urban area of Cape Town, South Africa, which presents complex social and environmental challenges. For each street-and-crossing segment, a virtual pedestrian streetscape audit tool was used to collect micro- and mesoscale environmental indicators and assess the quality of public space. This composite street-level assessment tool was weighted with a space syntax analysis indicator (i.e., spatial integration) to detect the network’s most interconnected and high-priority pathways. The Jenks natural breaks classification algorithm was used to classify scores for each segment, which ultimately found that the highest-priority streets for redevelopment are clustered in Bo-Kaap, a relatively disadvantaged, multicultural, and hilly district on Cape Town’s west side. Policy recommendations are evaluated to increase the quality of the urban environment and the city’s overall attractiveness to pedestrians. The proposed methodology facilitates more effective place management and classifies the city’s needs in improvements, minimizing both time and budget costs. [ABSTRACT FROM AUTHOR]

This study takes one step further to complement the application of a method for mapping informal green spaces (IGSs) using an efficient combination of open-source data with simple tools and algorithms. IGSs are unofficially recognized by the government as vegetation spaces designed for recreation, gardening, and forestry in urban areas. Due to the economic crisis, many formal green spaces such as urban parks and garden projects have been postponed, while IGSs have significant potential as green space retrofits. However, because they are small and spatially continuous and cannot be fully detected via airborne surveys, they are surveyed in small areas and neglected by government and city planners. Therefore, in this research, we combined the use of Google Street View (GSV) data with machine learning to develop a survey method that can be used to survey a wide area at once. Deeplab V3+ was used to segment the semantics based on the model created using 1000 labelled photos, with an accuracy rate of nearly 65%. Applying this method gave high accuracy in Ichikawa, Japan, with 3029 photos, and matched the results of a field survey in a previous study. In contrast, low accuracy was seen in Ho Chi Minh City, with 204 photos, where the quality of the GSV data was considerably lower. [ABSTRACT FROM AUTHOR]

Objective: While microscale pedestrian environment features such as sidewalks and crosswalks can affect pedestrian safety, it is challenging to assess microscale environment associated risk across locations or at scale. Addressing these challenges requires an efficient auditing protocol that can be used to assess frequencies of microscale environment features. For this reason, we developed an eight-item pedestrian environment virtual audit protocol and conducted a descriptive epidemiologic study of pedestrian injury in Washington State, USA.Methods: We used data from police reports at pedestrian-automotive collision sites where the pedestrian was seriously injured or died. At each collision site, high school students participating in an online summer internship program virtually audited Google Street View imagery to assess the presence of microscale pedestrian environment features such as crosswalks and streetlighting. We assessed inter-rater reliability using Cohen's kappa and explored prevalence of eight microscale environment features in relation to injury severity and municipal boundaries.Results: There were 2248 motor vehicle crashes eliciting police response and resulting in death or serious injury of a pedestrian in Washington State between January 1, 2015 and May 8, 2020. Of the crashes resulting in serious injury or death, 498 (22%) resulted in fatalities and 1840 (82%) occurred within municipal boundaries. Cohen's kappa scores for the eight pedestrian features that were audited ranged from 0.52 to 0.86. Audit results confirmed that features such as sidewalks and crosswalks were more common at collision sites within city limits.Conclusions: High school student volunteers with minimal training can reliably audit microscale pedestrian environments using limited resources. [ABSTRACT FROM AUTHOR]

Member states of the United Nations Convention to Combat Desertification are required to report on the proportion of land that is degraded in their countries, a requirement that is also tied into the UN Sustainable Development Goals (SDGs). National land degradation assessments are often conducted with the use of remote sensing data which are not always ground truthed. Google Street View (GSV) provides high resolution, panoramic imagery across large parts of the world that has the potential to be used to ground truth land degradation assessments. We apply three different methodologies (visual interpretation of GSV images, GSV image classification and vegetation index extraction) to derive vegetation cover estimates from Google Street View imagery for the Hardeveld bioregion of the Succulent Karoo biome in South Africa. Visual estimates of cover best predict known habitat condition values (adjusted R2 = 0.86), whilst estimates derived from an unsupervised classification of GSV images also predict habitat condition relatively well (adjusted R2 = 0.52). These results show the potential for using GSV imagery, and other large collections of ground-level landscape photographs, as a rough ground-truthing tool, especially in instances where more traditional ground-truthing approaches are not possible. [ABSTRACT FROM AUTHOR]