Trend AnalysisArts & Design

Street Art Documentation Through Computer Vision: Mapping Urban Creativity at Scale

Street art is inherently ephemeralโ€”murals are painted over, stickers peel, and buildings are demolished. Computer vision and street view imagery analysis now enable systematic documentation and spatial analysis of urban art at a scale impossible through traditional fieldwork.

By Sean K.S. Shin
This blog summarizes research trends based on published paper abstracts. Specific numbers or findings may contain inaccuracies. For scholarly rigor, always consult the original papers cited in each post.

Why It Matters

Street art occupies a paradoxical position in cultural documentation: it is among the most visible art forms (displayed on public surfaces for anyone to see) and among the most poorly documented (no institutional collection, no catalogue raisonne, no systematic archive). A mural painted today may be gone tomorrowโ€”painted over by property owners, weathered by climate, or demolished with its building. Google estimates that street art has an average lifespan of just a few years, yet it represents one of the most vibrant and culturally significant art movements of the 21st century.

Computer vision and street view imagery (SVI) offer a scalable solution. By training models to identify, classify, and geolocate street art in street-level imagery, researchers can create living maps of urban creativity that update as cities change. This intersection of art documentation and computational analysis opens new research possibilities: mapping the spatial distribution of artistic expression, tracking cultural influence patterns across neighborhoods, and preserving visual records of works that will inevitably disappear.

The Science / The Practice

Spatial Analysis of Graffiti in Urban Green Spaces

Cercleux et al. (2025) analyze graffiti and street art in Romanian urban parks, examining their spatial distribution and relationship to public discourse. The study uses systematic photographic documentation combined with spatial analysis to map where urban art appears in green spacesโ€”a context where art is often viewed as either enrichment or invasion. The research identifies patterns: street art clusters near park entrances and along major pathways, correlating with foot traffic rather than aesthetic considerations. This spatial intelligence has implications for both urban planning and cultural policy.

Street Art as Urban Transformation Agent

Pavel (2025) examines how street art and graffiti influence urban regeneration, community participation, and cultural identity in Bucharest. Using qualitative methods combining field observations and photographic documentation, the study demonstrates that street art functions as both a symptom and a driver of neighborhood transformation. Areas with high concentrations of street art attract creative economy workers, tourists, and media attentionโ€”creating feedback loops that accelerate gentrification. The documentation methodology itself is notable: systematic photographic surveys of entire neighborhoods over time, creating visual records of change.

Street View Imagery Mapping Frameworks

Ning et al. (2025), with 1 citation, introduce SIM, a general mapping framework for built environment auditing based on street view imagery. While not specifically focused on street art, SIM provides the technical infrastructure that enables large-scale visual documentation of any urban feature, including art. The framework demonstrates how Google Street View and similar platforms can be systematically analyzed using computer vision to create comprehensive spatial databases of urban visual elementsโ€”an approach directly applicable to street art mapping.

Computer Vision for Urban Visual Analysis

Ye et al. (2025), with 6 citations, review the use of street view imagery in urban analysis more broadly, demonstrating the maturity of computer vision techniques for extracting structured information from urban photographs. Their review catalogs the deep learning architectures (object detection, semantic segmentation, scene classification) that enable automated analysis of street-level imagery. These same techniquesโ€”trained on appropriate datasetsโ€”can identify murals, graffiti tags, stencil art, and paste-ups in millions of street view images, enabling documentation at a scale no team of human researchers could achieve.

Street Art Documentation Methods

<
MethodScalePrecisionTemporal CoverageLimitation
Manual fieldwork photographySingle neighborhoodHigh (context captured)Point-in-timeLabor-intensive, non-scalable
Street view imagery + CVCity or region-wideModerateMulti-year (SVI updates)Dependent on SVI coverage
Crowdsourced platforms (e.g., Street Art Cities)GlobalVariableOngoingBiased toward popular works
Drone surveyDistrict-levelHigh (multiple angles)Point-in-timeRegulatory constraints
Satellite imageryCity-wideLow (only large murals)Regular intervalsResolution limits

What To Watch

The convergence of computer vision, street view imagery, and cultural analytics will enable "living maps" of urban art that track creation, modification, and disappearance of works over time. Watch for AI models specifically trained to distinguish street art styles (stencil, wildstyle graffiti, mural, paste-up), enabling automated art-historical analysis of urban visual culture. The ethical dimensionโ€”whether systematic documentation of illegal graffiti aids or hinders artistsโ€”will require careful negotiation between researchers, artists, and authorities.

Explore related work through ORAA ResearchBrain.

References (4)

[1] Cercleux, A.-L., Banica, A., & Bogan, E. (2025). Decoding Graffiti and Street Art Attributes in Romanian Urban Parks: Spatial Distribution and Public Discourse. Sustainability, 17(12).
[2] Pavel, G. (2025). Painting the City: The Role of Street Art and Graffiti in Bucharest's Urban Transformation. Annals of the University of Bucharest - Geography, 74(1).
[3] Ning, H., Li, Z., & Yu, M. (2025). SIM: A mapping framework for built environment auditing based on street view imagery. arXiv.
[4] Ye, X., Li, S., & Gong, W. (2025). Street View Imagery in Traffic Crash and Road Safety Analysis: A Review. Applied Spatial Analysis and Policy.

Explore this topic deeper

Search 290M+ papers, detect research gaps, and find what hasn't been studied yet.

Click to remove unwanted keywords

Search 7 keywords โ†’