Deep Learning for Automatically Detecting Sidewalk Accessibility Problems Using Streetscape Imagery

Weld, Galen; Jang, Esther; Li, Anthony; Zeng, Aileen; Heimerl, Kurtis; Froehlich, Jon E.

doi:10.1145/3308561.3353798

Cited by 50 publications

(25 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…WheelMaps [50] 14 , or the feature "Wheelchair-accessible Places" in Google Maps 15 . Several systems attempt to increase information about accessibility of the built environment by crowd-sourcing approaches [37,44,50,60,61,71].…”

Section: Usage Of Maps and Technical Aidsmentioning

confidence: 99%

“…OpenStreetMap (OSM), Google Street View, and other resources, these information have reliability and timeliness issues [15,28,29,70]. Even if there are recent approaches to automatically collect information about the accessibility of the built environment by video footage [37,60], analysing images online manually [61] or automatically [71], or by sensor data [44,60], there is far too little information about the accessibility of buildings [66].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Traveling More Independently: A Study on the Diverse Needs and Challenges of People with Visual or Mobility Impairments in Unfamiliar Indoor Environments

Müller

Engel

Loitsch

et al. 2022

ACM Trans. Access. Comput.

View full text Add to dashboard Cite

It is much more difficult for people with visual or mobility impairments to prepare for a trip or visit unfamiliar places than it is for people without disabilities. In addition to the usual travel arrangements, one needs to know if the various parts of the travel chain are accessible. To the best of our knowledge, there is no previous work that examines in depth travel behaviour for indoor environments for both trip planning and execution, highlighting the special needs of people with low vision, blindness or mobility impairments. In this paper, we present a survey with 125 participants with blindness, low vision and mobility impairments. We investigate how mobile they are, what strategies they use to prepare a journey to an unknown building, how they orient themselves there and what materials they use. For all three groups, our results provide insights into the problem space of the specific information needs when planning and executing a trip. We found that most of our participants have specific mobility problems depending on their disability. Feedback from the participants reveals there is a large information gap, especially for orientation in buildings, regarding availability of high-quality digital, tactile and printable indoor maps, accessibility of buildings and mobility supporting systems. In particular, there is a lack of available and high-quality indoor maps. Our analysis also points out that the specific needs differ for the three groups. Besides the expected between-group differences, also large in-group differences can be found. The current paper is an expanded version of [18] augmented by data of people with mobility impairments.

show abstract

Section: Usage Of Maps and Technical Aidsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Traveling More Independently: A Study on the Diverse Needs and Challenges of People with Visual or Mobility Impairments in Unfamiliar Indoor Environments

Müller

Engel

Loitsch

et al. 2022

ACM Trans. Access. Comput.

View full text Add to dashboard Cite

show abstract

“…Recently, deep learning has been applied to Gooogle Street View images in order to detect accessibility problems (e.g., damaged sidewalks or obstructions) [29]. The main limitation of these techniques is that there are some features (e.g., a ramp inclination) that can be hard to detect with computer vision but that our inertial approach can indeed detect.…”

Section: Related Workmentioning

confidence: 99%

SmartWheels: Detecting urban features for wheelchair users’ navigation

Mascetti

Civitarese

Malak

et al. 2020

Pervasive and Mobile Computing

View full text Add to dashboard Cite

People with mobility impairments have heterogeneous needs and abilities while moving in an urban environment and hence they require personalized navigation instructions. Providing these instructions requires the knowledge of urban features like curb ramps, steps or other obstacles along the way. Since these urban features are not available from maps and change in time, crowdsourcing this information from end-users is a scalable and promising solution. However, it is inconvenient for wheelchair users to input data while on the move. Hence, an automatic crowdsourcing mechanism is needed.In this contribution we present SmartWheels, a solution to detect urban features by analyzing inertial sensors data produced by wheelchair movements. Activity recognition techniques are used to process the sensors data stream. SmartWheels is evaluated on data collected from 17 real wheelchair users navigating in a controlled environment (10 users) and in-the-wild (7 users). Experimental results show that SmartWheels is a viable solution to detect urban features, in particular by applying specific strategies based on the confidence assigned to predictions by the classifier.

show abstract

“…Consequently, research into pedestrian access has focused on enumerating and visualizing a handful of infrastructure elements at city-scale [21], small-scale (neighborhood or smaller) evaluations that consider pedestrian diversity [22][23][24], or large-scale network models based on a monolithic description of the pedestrian experience [25]. However, emerging means to scalably collect detailed pedestrian network data, including crowdsourcing in OpenStreetMap [26] and computer vision approaches [27][28][29][30][31] suggest that large-scale pedestrian networks will become increasingly available resources for investigating complex urban research questions. In addition, the detailed modeling of pedestrian behaviors and needs is increasingly suitable for scientific test and evaluation through the use of location tracking technology [32][33][34][35] and detailed surveys [36,37].…”

Section: Introductionmentioning

confidence: 99%

Towards routine, city-scale accessibility metrics: Graph theoretic interpretations of pedestrian access using personalized pedestrian network analysis

Bolten

Caspi

2021

PLoS ONE

View full text Add to dashboard Cite

A wide range of analytical methods applied to urban systems address the modeling of pedestrian behavior. These include methods for multimodal trip service areas, access to businesses and public services, diverse metrics of “walkability”, and the interpretation of location data. Infrastructure performance metrics in particular are an increasingly important means by which to understand and provide services to an urbanizing population. In contrast to traditional one-size-fits all analyses of street networks, as more detailed pedestrian-specific transportation network data becomes available, the opportunity arises to model the pedestrian network in terms of individual experiences. Here, we present a formalized and city-scale framework, personalized pedestrian network analysis (PPNA), for embedding and retrieving pedestrian experiences. PPNA enables evaluation of new, detailed, and open pedestrian transportation network data using a quantitative parameterization of a pedestrian’s preferences and requirements, producing one or more weighted network(s) that provide a basis for posing varied urban pedestrian experience research questions, with four approaches provided as examples. We introduce normalized sidewalk reach (NSR), a walkshed-based metric of individual pedestrian access to the sidewalk network, and sidewalk reach quotient (SRQ), an estimate of inequity based on comparing the normalized sidewalk reach values for different pedestrian profiles at the same location. Next, we investigate a higher-order and combinatorial research question that enumerates pedestrian network-based amenity access between pedestrians. Finally, we present city-scale betweenness centrality calculations between unique pedestrian experiences, highlighting disagreement between pedestrians on the “importance” of various pedestrian network corridors. Taken together, this framework and examples represent a significant emerging opportunity to promote the embedding of more explicit and inclusive hypotheses of pedestrian experience into research on urban pedestrian accessibility, multimodal transportation modeling, urban network analysis, and a broader range of research questions.

show abstract

Deep Learning for Automatically Detecting Sidewalk Accessibility Problems Using Streetscape Imagery

Cited by 50 publications

References 41 publications

Traveling More Independently: A Study on the Diverse Needs and Challenges of People with Visual or Mobility Impairments in Unfamiliar Indoor Environments

Traveling More Independently: A Study on the Diverse Needs and Challenges of People with Visual or Mobility Impairments in Unfamiliar Indoor Environments

SmartWheels: Detecting urban features for wheelchair users’ navigation

Towards routine, city-scale accessibility metrics: Graph theoretic interpretations of pedestrian access using personalized pedestrian network analysis

Contact Info

Product

Resources

About