Home Index About Next Previous Comments

"Remote Realtor": Using Images in Publicly-Available Cameras for Commercial Real-Estate Value Assessment Peter Friedman Abstract    Choosing the location of a restaurant, bar, retail store, or other commercial outlet is often a time consuming task that relies on constant observation, and information gathered by querying neighborhood residents. This process requires a individual to be present at the target neighborhood which can be costly if the investor is not local. This work demonstrates a method for determining optimal locations for commercial establishments and assessing value to areas of real-estate using information gathered from publicly available webcams. Introduction    Publicly available webcams, can be used either in a manual (human viewer) or automated (computerized analysis of images received) system to determine optimal locations for commercial real-estate given parameters of type and number of potential customers. The work outlines a method, which can be automated, for determining optimal locations for businesses given such parameters. Methods    Taking images captured from a camera atop the University of Pittsburgh Cathedral of Learning (https://www.discover.pitt.edu/tour/cl_cam.html), traffic patterns throughout the course of two weeks is analyzed. The analysis concentrated on three locations on the Pitt campus: Forbes and Thackeray Ave, Forbes Avenue directly below the Cathedral, and the corner of Forbes and Oakland ave. Results    Location 1: Fifth and Thackeray intersection at night (above) and at afternoon (below) on weekdays. This particular intersection exhibits the behavior or being busy with cars in both day and evening hours, however is only populated heavily with pedestrians during day hours – If a commercial developer were looking to place a walk in store (boutique) or other similar establishment this type of demographic may be preferred based on the presence of people averaged over business hours. Location 2: Hillman Library The Hillman Library location exhibited substantial pedestrian traffic in both day and evening hours – because of the presence of a well-visited (as evidenced by the camera video) academic building, developers looking to establish school/office supply, or convenience store may consider this location. Location 3: Forbes and Oakland Ave Location 3 was populated by heavy automotive traffic during daylight hours but relatively low pedestrian traffic. However, nighttime pedestrian traffic was much higher. As evidenced by the pre-existing assortment of restaurants and bars in this area, commercial developers seeking to establish "night-life" may consider locations such as these more valuable. While the work herein provides a commercial application to research previously done by Dr Sweeney in a limited applicable environment (University cameras), there are many more possibilities to investigate. Similar applications to commercial real-estate seem possible using data from highway traffic cameras (e.g. best location for a fast-food turn-off), much more data needs to be collected and verified before other applications are deemed feasible and/or practical. Privacy    As with nearly all public webcams there are privacy concerns to address, notedly the ability of users with access to sufficiently high-resolution cameras to obtain information about the vehicles (e.g. license plates ) or the people walking on the streets., Dr Sweeney cited similar concerns in her previous works. A unique problem to webcams mounted on tall buildings with pan, zoom, and tilt, in urban areas, is the ability to look into residential apartment windows. While the Pitt camera did not have zoom or resolution subtable to accomplish such a task, higher-quality cameras could easily be misused to look into peoples' private living-spaces. References    Camera: https://www.discover.pitt.edu/tour/cl_cam.html Project Paper (PDF) Gross, R and Sweeney, L. Mining Images in Publicly-Available Cameras for Homeland Security. AAAI Spring Symposium, 2005. Sweeney, L. CameraWatch. Data Privacy Lab, 2003 dataprivacylab.org/dataprivacy/projects/camwatch/