Street View Observer
LLM | Python | Data Visualization
Independent Study
Instructor:
Kristen Kurland
Street View Observer is an exploratory tool that uses Google Street View imagery and large language models (LLMs) to document and interpret changes in the urban environment over time.
By pairing historical panoramas with AI-assisted visual analysis, the project transforms street-level images into structured observations about business activity, infrastructure condition, and environmental quality.
Instead of relying solely on field visits or periodic surveys, this approach treats Street View as a continuously updated urban record — enabling planners to “walk the street” remotely and systematically across years.
Potential Observations
Using street-level imagery as a continuous urban record, the Street View Observer identifies visible indicators of neighborhood change across multiple planning domains. From storefront turnover and pedestrian activity to pavement repairs, streetscape maintenance, and environmental conditions, the system translates everyday visual cues into structured observations. These signals help planners quickly screen large areas, detect change over time, and surface issues or opportunities that may otherwise require extensive fieldwork. Rather than replacing site visits, the tool expands the planner’s reach — enabling remote, repeatable, and scalable urban observation.
Method
The workflow combines geospatial filtering, image collection, and LLM-based interpretation.
Street View panoramas are first queried within a defined radius and organized by capture date. Selected image pairs are then compared using a multimodal language model to identify visible changes, which are translated into planning-relevant observations.
This pipeline converts raw imagery into readable insights — turning visual differences into actionable urban indicators.
Data: Google Street View API
Scale: 25,000+ panoramas within 0.5 km radius
Tools: Python · GIS · Multimodal LLMs · Image comparison