With smartphones now capturing the moment with live video streams, a New York-based startup wants to make it easier to find something to watch.
The team at Dextro unveiled software on Tuesday that helps discover publicly shared content from Twitter-owned Periscope, an app that lets people stream live video from their phones’ cameras.
Dextro’s CEO and co-founder David Luan says his startup’s technology can do more than make it easier to find videos of cats at play. “We want Dextro to be the general visual, data analytic layer for everything on the Internet and connected devices,” he says.
Dextro can catalogue live video for a variety of needs, he says, such as discovering things of interest, analyzing social media for trending topics, or even automating home monitoring. “It’s all the same shared platform, taking unstructured human-themed clutter, parsed into computer vision to figure out what’s going on,” Luan says.
As live video streams began to gain in popularity, Luan saw that much of the public content was not tagged. Putting Dextro’s technology to work brought to light footage that might not otherwise be found. “We picked up video after the Baltimore protests, and the aftershocks of those protests,” he says.
The startup’s platform highlights prominent video streams of the moment, Luan says. There is no search function, but users can click on subcategories such as dash cams, pets, or people playing music to find something of interest. “We show what people are doing across the world at any given time,” he says. Though Dextro currently only picks up content from Periscope, the API could handle other streaming services, he says.
The idea for Dextro started to take shape when Luan was living in San Francisco and he saw a trend in that city’s hardware startup scene. In particular, there was an uptick in cameras being built into robots and connected devices.
Luan says he and co-founder Sanchit Arora both studied computer vision for robotics, and worked on a simple API that connected devices could use to do “smart things” with what they saw. “We hacked together a primitive version of this to see what people want think,” Luan says. They got contacted by developers and large companies who wanted to use the API for a variety of things such as social media analytics or categorizing online content.
That led to a more robust effort, he says. “We spent a lot of time building our technology and developing our machine learning algorithms, specifically to work with the special type of data that is video.”
Dextro has so far raised $1.7 million from angel investors including Esther Dyson and the co-founders of Medidata and Skype, venture capital firms RRE Ventures and Two Sigma Ventures, and from Yale University.
Beyond making it easier for the layperson to find video streams of interest, Luan foresees an opportunity for companies and services with huge, live video data sets to use Dextro’s platform. For instance, brands and marketers might use it to see when their products appear in live video.
While Dextro might not allow for full searches currently, that could change quickly. “We can do that right now, but there doesn’t seem to be enough content yet on Periscope for us to make that a viable feature,” Luan says.
