"Where Can I Park?" Understanding Human Perspectives and Scalably Detecting Disability Parking from Aerial Imagery

TL;DR

The paper tackles the problem of understanding and auditing disability parking through a dual approach: qualitative insights from PwDs and a scalable computer vision pipeline, AccessParkCV, that detects and characterizes disability parking from high-resolution aerial imagery. It presents an open, labeled dataset with 11,762 objects across 7 classes from Seattle, DC, and Spring Hill, and demonstrates state-of-the-art detection (micro-F1=$0.89$) and width estimation accuracy (average width error=$5.40\%$). The work additionally demonstrates practical applications via a personalized parking search tool and an urban analytics visualization, grounded in user-centered design recommendations. It offers design guidelines for policy and tool development, discusses limitations (notably per-city generalization and limitations of orthorectified imagery), and outlines future directions including multi-temporal data and human-in-the-loop verification. Overall, the study advances both the empirical understanding of disability parking and the technical capability to audit and plan accessible parking at scale.

Abstract

Accessible parking is critical for people with disabilities (PwDs), allowing equitable access to destinations, independent mobility, and community participation. Despite mandates, there has been no large-scale investigation of the quality or allocation of disability parking in the US nor significant research on PwD perspectives and uses of disability parking. In this paper, we first present a semi-structured interview study with 11 PwDs to advance understanding of disability parking uses, concerns, and relevant technology tools. We find that PwDs often adapt to disability parking challenges according to their personal mobility needs and value reliable, real-time accessibility information. Informed by these findings, we then introduce a new deep learning pipeline, called AccessParkCV, and parking dataset for automatically detecting disability parking and inferring quality characteristics (e.g., width) from orthorectified aerial imagery. We achieve a micro-F1=0.89 and demonstrate how our pipeline can support new urban analytics and end-user tools. Together, we contribute new qualitative understandings of disability parking, a novel detection pipeline and open dataset, and design guidelines for future tools.

Figures (12)

• Figure 1: Three example design probes of the 13 total shared with participants: (a) a routing tool showing the path from a parking space to the entrance of a destination, estimating both driving time and time to traverse to the entrance, (b) information about a specific place of interest, including disability parking space counts , and (c) a settings page, where one can filter for disability parking based on personalized needs. See the supplementary materials for the full 13 probes.
• Figure 2: Participants placed decals, signs, and cones on or near their vehicles to warn other drivers not to obstruct the space needed for them or their wheelchair to enter and exit the vehicle. Photos from participants P7 and P2.
• Figure 3: Our dataset contains 11,762 objects in seven categories: access_aisle, curbside, dp_no_aisle, dp_one_aisle, dp_two_aisle, one_aisle, and two_aisle. The images show examples for each parking category, with access_aisle implicitly included in the latter four. We include typical and atypical examples, and examples of the resulting detections from our pipeline (with parking detections in cyan and aisle detections in orange).
• Figure 4: Examples of potential disability parking spaces that we decided not to label due to overly narrow aisles, spaces being blocked due to garbage containers or other obstructions, occlusions from trees or shadows, or generally poor definition images.
• Figure 5: Examples where the research team struggled to determine disability parking and/or their bounds. (a) A space with a logo, which is completely blocked by a temporary structure. (b) A space with non-visible lines. (c) An atypical striped zone painted on top of what may be a parking space. (d) The only case where striping overlaps the logo, creating ambiguity about whether it is a parking space.