PathwayBench: Assessing Routability of Pedestrian Pathway Networks Inferred from Multi-City Imagery

TL;DR

PathwayBench addresses the challenge of generating routable pedestrian pathway graphs from remote-sensing data by providing a multi-city dataset with ground-truth pedestrian graphs and a novel local traversability metric. The approach centers on polygon-based local analysis (TIP) to proxy global routability, enabling efficient evaluation of connectivity beyond pixel-level overlap. Key contributions include externally validated ground-truth graphs, segmentation-friendly annotations, and a traversability-based evaluation that better aligns with real-world routing needs than traditional edge-count metrics. Findings show that multi-source inputs improve segmentation while routability metrics reveal strengths and weaknesses in existing methods that pixel-wise metrics miss, supporting more application-relevant benchmarking and cross-city generalization.

Abstract

Applications to support pedestrian mobility in urban areas require a complete, and routable graph representation of the built environment. Globally available information, including aerial imagery provides a scalable source for constructing these path networks, but the associated learning problem is challenging: Relative to road network pathways, pedestrian network pathways are narrower, more frequently disconnected, often visually and materially variable in smaller areas, and their boundaries are broken up by driveway incursions, alleyways, marked or unmarked crossings through roadways. Existing algorithms to extract pedestrian pathway network graphs are inconsistently evaluated and tend to ignore routability, making it difficult to assess utility for mobility applications: Even if all path segments are available, discontinuities could dramatically and arbitrarily shift the overall path taken by a pedestrian. In this paper, we describe a first standard benchmark for the pedestrian pathway graph extraction problem, comprising the largest available dataset equipped with manually vetted ground truth annotations (covering $3,000 km^2$ land area in regions from 8 cities), and a family of evaluation metrics centering routability and downstream utility. By partitioning the data into polygons at the scale of individual intersections, we compute local routability as an efficient proxy for global routability. We consider multiple measures of polygon-level routability and compare predicted measures with ground truth to construct evaluation metrics. Using these metrics, we show that this benchmark can surface strengths and weaknesses of existing methods that are hidden by simple edge-counting metrics over single-region datasets used in prior work, representing a challenging, high-impact problem in computer vision and machine learning.

Figures (4)

• Figure 1: (a) PathwayBench evaluates solutions to the pedestrian pathway inference problem: Given a dataset of co-registered aerial images, street map images, and road networks (Section \ref{['sec:dataset']}), produce a routable pedestrian network graph. (b) A new metric of local traversability (Section \ref{['sec:tiletraversability']}) provides an efficient proxy for global graph routability that centers the traveler experience.
• Figure 2: Samples from the PathwayBench dataset: imagery and annotations from various geographic areas offer great coverage and diversity.
• Figure 3: Traversability measures the ability to navigate from one boundary of a polygon to another using the graph, favoring global routability over local variations. Boundary pair $g[4]$ and $g[3]$ is traversable; $g[1]$ and $g[2]$ are not.
• Figure 4: Qualitative assessment on segmentation. The model that uses both aerial satellite images and street map images generates better predictions than the models that use only one input.