Unpacking visual perception of urban public space using social media images and street view imagery: Insights from Singapore riverfront

Visual perception, increasingly accessible through social media and street view platforms, facilitates broad public engagement and supports evidence-based decision-making. Despite notable advancements in perceptual assessment enabled by computer vision technologies, challenges remain in integrating diverse visual datasets, evaluating urban design strategies, and incorporating heterogeneous public perspectives. This study examines visual perception in the linear public spaces along the Singapore Riverfront, using the Singapore Urban Redevelopment Authority's five urban design guidelines for public spaces as the analytical framework. Visual perception is examined through both subjective and objective lenses using geotagged images from Flickr and Google Street View. The methodology incorporates Cityscape-Riverscape-Landscape clustered image classification, panoptic segmentation, and spatial mapping analysis. The visual perception results are mapped with the multi-source objective data associated with the five urban public space design guidelines. Key findings include: (1) The Singapore Riverfront exemplifies effective implementation of five strategies: enhanced riverscape visibility of the three quays, provision of essential public amenities, presence of distinctive commercial features, availability of open spaces and parks, and maintaining high-quality greenery. (2) Two datasets reveal comparative insights: Flickr emphasizes visual attraction for Clarke Quay and Boat Quay, while Google Street View reveals the river's broader contributions to spatial diversity, space equity, and environmental sustainability. (3) Spatial mapping results inform data-driven placemaking to enhance the spatial quality and functionality of the Singapore Riverfront. These results can contribute to a human-centered design policy feedback loop and advance the evaluation of urban environments.