When choosing a centerline connected graph layer representation for the OpenSidewalks schema, one of the many motivations was the ability to construct a graph from simple observations and inputs. Further, in the case of the non-motorized transportation layer, the graphs allow as few or as many attribute overlays as the data producers are able to provide. This structure makes it easy for stakeholders to understand and visualize the network, regardless of their technical background.

• Mobile and Low-Tech Solutions: Basic GPS devices or smartphones can record paths and waypoints, which can be converted into graph data. We have created multiple applications for this type of collection.
• Community Participation: Local residents and volunteers can easily map out nodes and edges using simple tools like paper maps or mobile apps.
• Crowdsourcing: Use of crowdsourcing platforms (e.g., OpenStreetMap) to gather data from a broad user base, even those with minimal technical expertise.
• Computer-vision techniques: Advanced methods for data collection may leverage both aerial and street-level imagery. These methods can be particularly useful in low-resource settings due to their ability to automate and scale the data collection process. Here’s a detailed look at how computer-vision can be used:
  1. Aerial/Satellite Imagery (PROPHET- tinyurl.com/2023APE)
     • Data sources
       • Drones: Affordable drones equipped with cameras can capture high-resolution images of urban and rural areas.
       • Satellite Imagery: Freely available satellite images from sources like Google Earth or open data initiatives provide extensive coverage.
     • Techniques
       • Image Segmentation: Identify different types of surfaces (e.g., roads, sidewalks, bike lanes) by segmenting images into distinct regions.
       • Object Detection: Detect and classify objects like pedestrian crossings, bike racks, and traffic signs.
       • Pathway Extraction: Extract pathways by identifying linear features in the imagery that correspond to walking and biking routes.
     • Advantages
       • Wide Coverage: Aerial imagery can cover large areas quickly, making it suitable for city-wide or regional assessments.
       • Automated Analysis: Advanced algorithms can process vast amounts of imagery data, reducing the need for manual input.
       • Frequent Updates: Satellite imagery is frequently updated, allowing for the monitoring of changes over time.
  2. Street-Level Imagery
     • Data sources
       • Mobile Mapping through Wheelchairs and pushed strollers: Equipped with pedestrian-POV RGBD cameras, these hardware solutions can capture detailed street-level views.
       • Crowdsourced Data: Contributions from platforms like Google Street View and Mapillary, where users upload images captured via smartphones.
       • Fixed Cameras: Cameras installed on street infrastructure, such as traffic lights or buildings, provide continuous data feeds.
     • Techniques
       • Pedestrian and Cyclist Detection: Use computer vision to identify and count pedestrians and cyclists, helping to measure usage and demand.
       • Infrastructure Identification: Detect the presence and condition of sidewalks, bike lanes, benches, lighting, and other relevant infrastructure.
       • Accessibility Features: Identify accessibility features like ramps, tactile paving, and signalized crossings for disabled individuals.
     • Advantages
       • User Participation: Crowdsourced street-level imagery encourages community involvement and increases data granularity and coverage.
       • Detailed View: Street-level imagery provides a granular view of the infrastructure, enabling the identification of specific features and conditions.
       • Real-Time Data: Fixed cameras and mobile mapping can offer real-time or near-real-time data collection.