This startup raised $3.25 million for its AI video technology — now it's pivoting to tell shoppers whether a store is crowded before they arrive

  • AI startup Voxel51, which has raised $3.25 million, used its video annotation technology to gather physical distancing data once the pandemic hit.
  • The tool, called the Physical Distancing Index, can look through video footage to understand how many people are in a space and how they are moving through it.
  • Jason Corso, the CEO of Voxel51, said that his team can tailor this tool to create apps for retailers that would allow shoppers to remotely check how many people are at a given store in real-time.
  • Corso's hope is that retailers could use this technology to maintain safe distancing measures, reduce lines or traffic, and help consumers make informed decisions about whether to shop at a given time.
  • Visit Business Insider's homepage for more stories.

Standing in a socially-distanced line that snakes around your grocery store parking lot or down a long city block has become a commonplace experience for many since March. This stop-gap fix has worked fine enough, but as the US hits month 6 of living with coronavirus, longer-term solutions could be useful. Wouldn't it be great if before arriving at an over-crowded grocery store, you could check an app remotely to see how many people are in the store at that exact moment, and make an informed decision on whether or not to go? Jason Corso, co-founder and CEO of machine learning startup Voxel51, wants to help consumers do exactly that.

The team at Voxel51 builds machine-learning tools that enable new AI applications. Before the pandemic, they had created a suite of tools that are able to quickly recognize and annotate video footage by teaching AI to understand motion and identify cars, humans, and bicycles anonymously. The tools can also create visualizations of that data. This tech was traditionally used by municipalities, but when the pandemic hit, Corso started to think about how Voxel51 could help out.

"We wanted to mobilize. We asked ourselves what we could do to help," he told Business Insider in an interview. "The Physical Distancing Index came out of that question."

The Physical Distancing Index is Voxel51's new product that uses their suite of video annotating tools. The PDI takes data from video footage and visualizes how people move through a physical space. The tool, Corso explains, is open-source, which means that it's freely available on the internet (Github specifically) for other researchers or developers to take the technology they've created and apply it to a specific situation. 

Corso feels confident that he could work with retailers to make PDI-powered customer-facing apps that would allow shoppers to see how busy a store or parking lot is remotely, in near real-time (currently the app refreshes every 15 seconds, but Corso says it could be changed to refresh more frequently). With such accurate data, shoppers would be able to make safer decisions for themselves based on their vulnerabilities and concerns around COVID-19.

The closest existing tool might be the function on Google Maps that uses historic data to project how busy a store typically is at that time. But the real-time aspect of PDI gives shoppers more confidence.

"With something like PDI connected to a store you could get a more real-time understanding of what's happening right now," he said.

From a customer standpoint, this saves people from getting in their cars or schlepping by foot or public transit to their grocery store, only to arrive and see the line is outlandish or the parking lot is packed. From a retailer standpoint, giving customers the power to see how busy a store is means that traffic spikes could be smoothed out by consumers who are self-selecting away from busy spurts. Retailers could also better manage the flow of traffic through their stores using data provided by the PDI.

Voxel51 has also just released a new tool, called FiftyOne, that allows users of their machine learning tool to customize it for their uses even more deeply. For example, he said you could teach the AI to recognize not just humans and vehicles, but also dogs. Imagine knowing how many dogs are at the dog park, so you can ensure your pup will have a playmate before you arrive.

With some experts predicting the coronavirus might never disappear, even after a vaccine, creating a customer-facing app using the PDI technology could be a good investment that helps everyone adjust to the new normal of shopping.

"Trader Joe's might be doing fine now with that line out the door because it works right now and it's not an expensive solution for them," Corso said. The problem, he explained, is that we don't know what the future holds. Retailers could be reticent to invest in creating better solutions to these problems because they're unsure how long these problems will last. But COVID-19 has already had major effects on shoppers' behaviors that are predicted to last beyond the pandemic. And being able to know what your shopping experience will look like before you go is a benefit shoppers could get used to.

Source: Read Full Article