StoreFront Visibility

Inspiration

Store visibility is an important consideration in retail location decisions because it directly impacts customer traffic and, consequently, revenue. A location with high visibility, such as a corner spot or a storefront facing a busy street, ensures that potential customers can easily see and access the business. Conversely, a hidden or obscured location can drastically limit foot traffic, even if the surrounding area has high population density. Effective visibility translates to increased brand awareness, spontaneous purchases, and ultimately, a greater chance of business success. However, accurately quantifying a location's visibility is a complex task, as it involves factors beyond simple street frontage, such as pedestrian flow patterns, competing signage, and the presence of visual obstructions, making it difficult to obtain reliable data and make informed decisions.

What it does

Our solution determines the visibility of buildings by analyzing two key factors: 1) The area of the building that is obstructed by objects in the foreground. 2) The traffic volume on the street from which the building is being observed.

Using Google Maps API, OpenStreetMap (OSM), and Meta’s Detectron2, we assess the visibility of storefronts from various perspectives within a 75-meter radius. The final output is a visibility score that businesses can use to evaluate how effectively their storefronts are seen by pedestrians and drivers.

How we built it

1) Address Input & Data Retrieval Users enter a specific address. We use GrowthFactor.AI’s dataset to identify streets within a 75-meter radius. 2) Street View Image Collection We obtain street view images using latitude and longitude values from OSM and the Google Maps API. 3) Object Detection & Visibility Mapping We process each image through Detectron2 (using Mask R-CNN for panoptic segmentation) to identify and segment buildings and foreground objects. The building’s visible area is calculated as a fraction of the total building area. 4) Traffic Data Integration We retrieve traffic volume data for the corresponding streets using OSM. The final visibility score is computed by multiplying the visible area ratio by the traffic volume. 5) Output & Insights The system provides a visibility score that businesses can use for marketing strategies and location optimization.

Challenges we ran into

Accuracy of Building Detection: Detectron2’s masks are not always perfectly precise, affecting visibility calculations. Computational Limits: Due to GPU constraints on Google Colab, we had to limit image processing iterations. Perspective Assumptions: Our model assumes that images captured from street views are representative of both pedestrian and driver visibility, but it does not account for sidewalk positions or varying vantage points.

Accomplishments that we're proud of

Successfully integrating Google Maps API, OSM, and Detectron2 into a cohesive solution. Developing a scalable approach that can be applied to storefronts globally. Creating a practical tool that provides businesses with valuable insights into their visibility.

What we learned

The importance of precise object detection in real-world applications. How to efficiently handle and integrate multiple datasets. The challenges of computational limitations when working with large-scale image processing.

What's next for StoreFront Visibility

Enhancing Object Detection: Improving Detectron2’s accuracy to better distinguish between different building types. Pedestrian vs. Driver Perspective Analysis: Incorporating additional data to differentiate visibility based on pedestrian walkways and driving lanes. Weather & Time Factor Integration: Accounting for how rain, snow, and nighttime conditions impact storefront visibility. Expanded Dataset & Locations: Scaling the project to analyze more addresses worldwide, increasing its applicability for businesses seeking prime locations.

Built With

• python