Inspiration
Store visibility is a crucial factor for business success, influencing foot traffic and revenue. However, accurately quantifying visibility remains a challenge. We aimed to develop a data-driven approach to measure and enhance store visibility using geospatial data, and cloud APIs. Our goal was to create a tool that helps businesses optimize their location visibility with actionable insights.
What it does
StoreScope evaluates the visibility of a storefront by analyzing its distance from the nearest road, its angle of view computed as a bearing from the road, the presence of obstructing objects and the volume of traffic in that road segment. It gathers geospatial data from Google’s Geocoding API and Google’s Road API, retrieves images via Google’s Street View API, and uses Gemini’s Cloud Vision API to assess obstructions. The Overpass API is used to cross-reference OpenStreetMap (OSM) data and query the provided database. Using mathematical modeling, the tool calculates and normalizes a visibility score that businesses can use to optimize their storefront presentation and location strategy.
How we built it
Data Collection & Preprocessing: Geocoded addresses into latitude and longitude. Retrieved road proximity using Google’s Road API. Extracted storefront images via Google’s Street View API. Processed obstruction data using Gemini’s Cloud Vision API. Cross-referenced OSM indexes using the Overpass API. Cleaned and optimized data by removing unnecessary variables and merging redundant entries.
Mathematical Modeling: Visibility score V was formulated as a function of: D (distance from road) θ (angle of view) O (obstructing objects) An exponential decay function was used to model distance attenuation: VD = e^(-αD) * V (where α is the attenuation coefficient, set to 0.1 for consistency) The cosine squared of the angle was incorporated to account for its effect on visibility. Obstruction percentage was derived from image analysis. The final visibility score was normalized between 0 and 1 and displayed as a percentage.
Challenges we ran into
Data Variability: Address formats in Google Maps were inconsistent, affecting nearest road calculations. Street View Accuracy: Some images were outdated or did not accurately represent the storefront. Corner Stores: Stores with multiple street-facing sides posed an issue, as only one was analyzed. Computational Efficiency: Processing large datasets was resource-intensive, requiring careful optimization.
Accomplishments that we're proud of
Successfully integrated multiple APIs to create a functional, automated visibility evaluation system. Developed a robust mathematical model that accurately quantifies storefront visibility. Optimized data preprocessing to improve performance and scalability. Demonstrated the real-world applicability of our approach for businesses looking to optimize store placement.
What we learned
Importance of Data Cleaning: Inconsistent and redundant data significantly affect model accuracy and computational efficiency. API Limitations: Understanding and working around API constraints was crucial in ensuring reliability. Scalability Considerations: While our approach is scalable, incorporating machine learning can enhance accuracy and adaptability.
What's next for StoreScope
Storescope's prediction can be made more accurate with larger data sets and using machine learning models that account for other factors in visibility.
Log in or sign up for Devpost to join the conversation.