With exaggerated hand gestures, data science intern Aldebaran Saldaña comedically ushers me in and out of a life-size “elevator simulator” — a small, curtained-off rectangle with a monitor mounted in the top right corner. “I feel like a border security officer,” he says. “Passports, please!”
Rishabh Shukla, another intern, shares the system’s findings from behind a cluster of monitors. “It estimates that you are a 27-year-old male.”
While I am actually a 29-year-old male, the interns remain unfazed by the software’s slight underestimation of my age. Given my extensive skin care routine and impeccable grooming, it is an understandable mistake. Enormous advancements will have to be made in the field of AI before machines can successfully guess my age.
I am at the main offices of Vertical City, an Edmonton-based advertising technology company specializing in digital elevator screens. If you live or work in a multi-story building, it is likely you have come across screens akin to those deployed by Vertical City. Usually mounted in the corners of elevators, these small displays exhibit diverse content — including advertisements, weather forecasts, news updates, building announcements, and more — offering passengers information and entertainment during their vertical travels.
However, Vertical City’s screens do more than just display content. Equipped with cameras and specialized software, these screens measure the impact of the advertisements and bulletins they exhibit, producing performance reports for clients. These performance reports include measures such as audience reached and the attention shown to the screen content.
Notably, this technology generates purely anonymous data that can’t be linked to specific individuals. It holds certification from the Privacy by Design Centre of Excellence, an internationally renowned standard for privacy compliance.
Growth through data
To gain a more comprehensive understanding of its data, Vertical City partnered with the Applied Data Science Lab (ADSL) this past summer. This Cybera-run program empowers companies to experiment with their data by providing them with teams of data scientists. These teams comprise master’s students, recent university graduates, and former engineers seeking to establish themselves in the burgeoning new field. Guiding and mentoring of each team is done by a seasoned data science professional from Cybera.
Industry partners make a modest upfront contribution to demonstrate their commitment to the program, while Cybera provides training and assumes the cost of labour. This arrangement makes data exploration initiatives significantly more feasible, particularly for risk-averse small to medium-sized businesses.
“We had an overwhelming amount of data, but lacked the time and resources to explore it thoroughly,” says Derek Moody, VP Insights for Vertical City. “The interns from Cybera, with the support of their mentor, are conducting scientific experiments to further dissect our data and uncover its latent possibilities. Thanks to them, we can elevate our data science initiative to new heights.”
Understanding the audience
On the day I visited, the team was conducting one such experiment. “Our current focus is on analyzing the system’s accuracy in estimating the age and gender of passengers,” explains data science intern Chinwe Ajieh.
The testing process was straightforward. A selected volunteer stepped into the test elevator and remained inside for a specified period. In each test, the team manipulated certain variables, such as the volunteer’s positioning and direction within the elevator. Additionally, some volunteers were asked to wear specific accessories like hats, sunglasses, hoods, and other adornments. After the designated time elapsed, the volunteer stepped out of the elevator, and the team compared the system’s estimations of his or her age and gender with their actual stated age and gender.
As I was arriving, the interns were searching for an exceptionally handsome man in his late 20s for the experiment. Upon witnessing my impeccable visage, they promptly invited me to take part. I graciously accepted this invitation.
But I’m just one volunteer. Over the course of the day, a diverse range of individuals entered and exited the simulator, all under the guidance of Aldebaran’s dramatic gestures and playful seriousness.
While this type of exploratory work might be time-consuming and sometimes tedious, it often yields actionable insights, expediting product improvement. According to Jordan Swanson, the team’s data science mentor, those trials prompted several software optimizations.
“The software gauges passengers’ age and gender by analyzing the camera’s video stream. That day’s tests have already revealed some frame combinations are better suited for age detection, while others are better suited for gender detection. As a result, we refined the algorithm. This modification will improve the software’s precision.”
Beyond sparking cost-saving and product-improving tweaks and refinements, these insights shape cutting-edge AI tools.
“Exploratory data analysis is an essential — albeit demanding — phase in any AI-focused project,” says Shivam Rathore, another member of the data science team. “Our efforts are poised to establish the groundwork for forthcoming machine learning endeavors that will propel Vertical City to the forefront of advertising technology.”
How to get involved
For more information on who is eligible to join a future cohort of the Applied Data Science Lab — either as an industry partner, or an intern — please check out our webpage.