ANNUAL CONGRESS 2025

An Integrated Visual Framework: Using SVIs to Inform Detailed Street Renewal in Historic Urban Spaces

Not scheduled

20m

Yildiz Technical University, Istanbul

Oral Track 06 | URBAN CULTURES AND LIVED HERITAGE

Speaker

Mr Haoyang He (College of Architecture and Urban Planning, Tongji University, Shanghai 200092, China)

Description

Background: In the context of increasingly unsettled urban development, historic urban spaces serve as crucial physical carriers of a city's culture and memory, providing collective connections that shape the city's sense of social identity. These visual elements of historic streets constitute spatial memory patterns that are deeply intertwined with people's cultural lives. However, with the rapid development of cities, historic urban areas are facing ageing challenges which increasingly demand detailed street renewal to address the various issues of urban instability and imbalance.
Research Gap: Due to the varying levels and unique cultural landscapes across different cities, the overarching policies of core cities are not often applicable to smaller towns. As a result, governments and stakeholders lack effective quantitative references for different scales of urban renewal, hindering the development of their own tailored strategies for cultural landscape protection. To solve this problem, most local governments conduct subjective analyses based on comprehensive local data of the city's unique cultural characteristics and detailed manual investigations for small-scale validation. However, in areas with limited data, the cost of manual investigations is high and the policies may lack quantitative measurements. Furthermore, limited investigations cannot collect data targeted on the specific renewal needs, making it difficult to obtain effective references from existing historic urban renewal cases.
Method: In the era of big data, urban street-view imagery (SVI) provides an opportunity for efficient quantitative analysis (Fan et al., 2023), offering rich visual information which includes diverse spatial elements, building colours, architectural styles, materials, and more (Zhou et al., 2023). These indicators, which support a human-centred analysis of historic streets, can directly inform detailed urban planning guidelines, especially for cultural landscape preservation (Chen et al., 2024). Our research selected Kaifeng as a case study and employed an integrated visual analysis framework to process SVI data for the renewal of historic urban spaces. In the analysis part, we first employed semantic segmentation via computer vision models to extract the spatial perception elements from historic streets of the city, including street vegetation, street building, and street enclosure. Subsequently, we isolated the street buildings in SVI pictures through the Deeplab models, applied computer vision to evaluate their colour distribution, and constructed a colour benchmark for the historical urban area. Finally, we trained a deep learning model for building material recognition and calibrated each historic street with its corresponding symbolic types. These three steps offer quantified references for street element renewal, colour harmony control, and material enhancement in historic urban spaces. In the following implementation, we applied our framework to exemplary cases from the cities of Xi’an and Luoyang, whose historic areas share similarities with Kaifeng, in order to compare the results and derive references that can directly inform detailed street renewal policies.
Conclusion: The quantified results of these indicators validated the efficiency of our SVI-based integrated visual framework, highlighting its capability for large-scale, high-precision measurement of historic urban places. By expediting the formulation of spatial optimization policies for urban renewal and cultural heritage preservation, this data-driven approach provides a computational design solution to address the challenges of dynamic and inequitable urban development.

References

Chen, X., Ding, X. and Ye, Y. (2024) ‘Mapping sense of place as a measurable urban identity: Using street view images and machine learning to identify building façade materials’, Environment and Planning B: Urban Analytics and City Science, p. 23998083241279992.
Fan, Z. et al. (2023) ‘Urban visual intelligence: Uncovering hidden city profiles with street view images’, Proceedings of the National Academy of Sciences, 120(27), p. e2220417120.
Zhou, Z. et al. (2023) ‘Evaluating building color harmoniousness in a historic district intelligently: An algorithm-driven approach using street-view images’, ENVIRONMENT AND PLANNING B-URBAN ANALYTICS AND CITY SCIENCE, 50(7), pp. 1838–1857.